KINGY.AI DEEPSEEK V4 FLASH 0731 — VERIFIED TWO-QUANT APPENDIX

PUBLICATION BOUNDARY
This appendix covers complete base runs for UD-IQ2_M and UD-IQ3_XXS only.
UD-IQ1_S and UD-Q4_K_XL were not recovered as locally verifiable completed runs.
UD-Q8_K_XL is represented only by a generation-free load/capacity smoke.
No five-quant quality winner is claimed. Long-context retrieval was NOT RUN.
Multilingual rows are literal/canonical-marker diagnostics only. They are excluded
from quality composites, rankings, recommendations, and headline claims. This
diagnostic does NOT ESTABLISH multilingual factual preservation, multilingual
degradation, or general translation fluency.

suite_id: kingy-deepseek-v4-flash-0731-cross-quant-v1
plan_id: cd6b4c0a2a9329ff395278dcc2733641212a0a70d6d9dca3660bb670f734f760
seeds: 731, 1731, 2731
context allocation: 32768 tokens
runtime: llama.cpp b10219 / c629da565c80b0b17fac6262acdca4d772e745d8
host: six NVIDIA A40 GPUs; 540,644,073,472 bytes physical RAM; 96 logical CPUs
benchmark threads: 52

LOCAL VERIFICATION
verification_file: work/runpod-live/full-run-jsqxbdwx8mi8ti/incremental/iq2-pull-verification.json
verification_sha256: 3b1fcfdc56518fc74fd32b6b5a6c8bc45b691ffc343b4e4fc8dcdb260b97c030
verified_manifests: 3
verified_base_labels: UD-IQ2_M, UD-IQ3_XXS
verified_smoke_labels: UD-Q8_K_XL

QUANT UD-IQ2_M
manifest: work/runpod-live/full-run-jsqxbdwx8mi8ti/pulled/evidence/quants/UD-IQ2_M/base/evidence-manifest.json
manifest_sha256: 8d91bf7413cebe711005431af7ea231961c3891ae4d8335cb847b62c186f0fd5
completed_utc: 2026-08-02T03:17:52.182235+00:00
executed_rows: 177
decision_included_rows: 159
multilingual_diagnostic_rows: 18
decision_score: 318/426
decision_percent: 74.647887
task_summary_json:
{"multilingual": {"included": 18.0, "max_score": 180.0, "score": 97.0}, "python_patch": {"included": 3.0, "max_score": 3.0, "score": 0.0}, "reasoning": {"included": 90.0, "max_score": 90.0, "score": 67.0}, "strict_json": {"included": 3.0, "max_score": 30.0, "score": 20.0}, "tool_calling": {"included": 60.0, "max_score": 300.0, "score": 231.0}, "typescript_patch": {"included": 3.0, "max_score": 3.0, "score": 0.0}}
performance_summary_json:
{"decode_tokens_per_second": {"maximum": 36.19035377318757, "median": 34.886633431697284, "minimum": 32.76015040588155}, "peak_process_rss_bytes": {"maximum": 6819237888.0, "median": 5516816384.0, "minimum": 1970454528.0}, "peak_system_memory_used_bytes": {"maximum": 56259399680.0, "median": 45963157504.0, "minimum": 40017698816.0}, "prompt_tokens_per_second": {"maximum": 750.0416531850613, "median": 336.59222204383303, "minimum": 218.15857488258345}, "swap_delta_bytes": {"maximum": 0.0, "median": 0.0, "minimum": 0.0}, "time_to_first_token_seconds": {"maximum": 1.853359289932996, "median": 0.7640615419950336, "minimum": 0.5134036749368533}}

ROW-LEVEL RECORDS
{"case_id": "LANG-de", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-de/seed-1731/checkpoint.json", "checkpoint_sha256": "d812891a84071ad006b06e88cdf474d2f50bbcb6a58d131828bfc60661cc7685", "completed_utc": "2026-08-02T02:14:52.588717+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 907, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.002741559824045, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 907, "prompt_tokens_per_second": 309.476078681829, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2014679040, "peak_system_memory_used_bytes": 40095485952, "per_gpu": {"0": {"peak_memory_used_mib": 17589.0, "peak_power_draw_w": 150.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16537.0, "peak_power_draw_w": 152.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16507.0, "peak_power_draw_w": 146.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19077.0, "peak_power_draw_w": 147.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16537.0, "peak_power_draw_w": 147.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15119.0, "peak_power_draw_w": 147.06, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 24, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5644251039484516}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 2.0, "semantic_adequacy": 2.0, "shipment_identifier": 1.0}, "language": "de", "max_score": 10.0, "passed": true, "score": 10.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 159, "path": "results/quants/UD-IQ2_M/cases/LANG-de/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "47d4659d1a744d487884d87407016f79f8fc69ad42c7f661d693a487354e24d6", "utf8": "Mara Chen hat Sendung VX-44 nicht genehmigt. Sie bleibt bis zum 17 August 2026 zurückgestellt. Das genehmigte Limit beträgt €12,450.75, nicht €12,540.75."}]}
{"case_id": "LANG-de", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-de/seed-2731/checkpoint.json", "checkpoint_sha256": "7a80d238bbc3a27511e32c49c727d161c1cbb79db6275ef465cdc0a9152bade2", "completed_utc": "2026-08-02T02:15:22.861877+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 1024, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.9061484794384, "finish_reason": "length", "output_token_cap": 1024, "output_tokens": 1024, "prompt_tokens_per_second": 310.25767196080875, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2058039296, "peak_system_memory_used_bytes": 41143562240, "per_gpu": {"0": {"peak_memory_used_mib": 17856.0, "peak_power_draw_w": 152.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16804.0, "peak_power_draw_w": 144.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16774.0, "peak_power_draw_w": 140.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19344.0, "peak_power_draw_w": 143.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16804.0, "peak_power_draw_w": 145.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15386.0, "peak_power_draw_w": 144.12, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 27, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5667027320014313}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 0.0, "hold_status": 0.0, "name": 0.0, "negation": 0.0, "semantic_adequacy": 0.0, "shipment_identifier": 0.0}, "language": "de", "max_score": 10.0, "passed": false, "score": 0.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 1, "path": "results/quants/UD-IQ2_M/cases/LANG-de/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "08f271887ce94707da822d5263bae19d5519cb3614e0daedc4c7ce5dab7473f1", "utf8": "M"}]}
{"case_id": "LANG-de", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-de/seed-731/checkpoint.json", "checkpoint_sha256": "b4deec1403a95bba96b0880ac55a51dcf8fcbbd5c272a1f8f9b04b1748072799", "completed_utc": "2026-08-02T02:14:25.756007+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 1024, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.78702797162411, "finish_reason": "length", "output_token_cap": 1024, "output_tokens": 1024, "prompt_tokens_per_second": 218.15857488258345, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 1970454528, "peak_system_memory_used_bytes": 40140136448, "per_gpu": {"0": {"peak_memory_used_mib": 17589.0, "peak_power_draw_w": 144.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16537.0, "peak_power_draw_w": 142.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16507.0, "peak_power_draw_w": 137.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19077.0, "peak_power_draw_w": 151.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16537.0, "peak_power_draw_w": 149.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15119.0, "peak_power_draw_w": 145.75, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 27, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7448726140428334}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 0.0, "hold_status": 0.0, "name": 0.0, "negation": 0.0, "semantic_adequacy": 0.0, "shipment_identifier": 0.0}, "language": "de", "max_score": 10.0, "passed": false, "score": 0.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/LANG-de/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "LANG-en", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-en/seed-1731/checkpoint.json", "checkpoint_sha256": "4ca1af894706977bfa895d7c7674a91cf15b4d7bf419ae953e98aca7ee23e267", "completed_utc": "2026-08-02T02:15:29.367182+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 58, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 36.19035377318757, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 58, "prompt_tokens_per_second": 312.86571130415496, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2102456320, "peak_system_memory_used_bytes": 40017698816, "per_gpu": {"0": {"peak_memory_used_mib": 17589.0, "peak_power_draw_w": 148.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16537.0, "peak_power_draw_w": 141.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16507.0, "peak_power_draw_w": 142.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19077.0, "peak_power_draw_w": 138.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16537.0, "peak_power_draw_w": 137.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15119.0, "peak_power_draw_w": 144.74, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 2, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5134036749368533}, "quant": "UD-IQ2_M", "score": {"canonical_exact": true, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 2.0, "semantic_adequacy": 2.0, "shipment_identifier": 1.0}, "language": "en", "max_score": 10.0, "passed": true, "score": 10.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 136, "path": "results/quants/UD-IQ2_M/cases/LANG-en/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "c3206d9f219883f5fee0a386cc815d5ca321881c6b6b0336ef4a9e9f98a1b9aa", "utf8": "Mara Chen did not approve shipment VX-44. It remains on hold until 17 August 2026. The approved limit is €12,450.75, not €12,540.75."}]}
{"case_id": "LANG-en", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-en/seed-2731/checkpoint.json", "checkpoint_sha256": "e7073931100808e17d8674391b1cc842a64b88fbb1c9b30065bcf96cf70adb2b", "completed_utc": "2026-08-02T02:15:37.615458+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 260, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.36588318539966, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 260, "prompt_tokens_per_second": 312.6614015542042, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2140413952, "peak_system_memory_used_bytes": 40147771392, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 147.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 142.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 137.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 141.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 140.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 140.61, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5214025300228968}, "quant": "UD-IQ2_M", "score": {"canonical_exact": true, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 2.0, "semantic_adequacy": 2.0, "shipment_identifier": 1.0}, "language": "en", "max_score": 10.0, "passed": true, "score": 10.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 136, "path": "results/quants/UD-IQ2_M/cases/LANG-en/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "c3206d9f219883f5fee0a386cc815d5ca321881c6b6b0336ef4a9e9f98a1b9aa", "utf8": "Mara Chen did not approve shipment VX-44. It remains on hold until 17 August 2026. The approved limit is €12,450.75, not €12,540.75."}]}
{"case_id": "LANG-en", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-en/seed-731/checkpoint.json", "checkpoint_sha256": "9c039e741598490075bb64372d930956ccb6821ad1b3ac94895cab4140ef55b1", "completed_utc": "2026-08-02T02:15:26.926133+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 112, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.686883286048214, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 112, "prompt_tokens_per_second": 313.0721492601194, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2102157312, "peak_system_memory_used_bytes": 40051757056, "per_gpu": {"0": {"peak_memory_used_mib": 17589.0, "peak_power_draw_w": 146.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16537.0, "peak_power_draw_w": 143.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16507.0, "peak_power_draw_w": 137.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19077.0, "peak_power_draw_w": 141.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16537.0, "peak_power_draw_w": 139.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15119.0, "peak_power_draw_w": 138.67, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5635613249614835}, "quant": "UD-IQ2_M", "score": {"canonical_exact": true, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 2.0, "semantic_adequacy": 2.0, "shipment_identifier": 1.0}, "language": "en", "max_score": 10.0, "passed": true, "score": 10.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 136, "path": "results/quants/UD-IQ2_M/cases/LANG-en/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "c3206d9f219883f5fee0a386cc815d5ca321881c6b6b0336ef4a9e9f98a1b9aa", "utf8": "Mara Chen did not approve shipment VX-44. It remains on hold until 17 August 2026. The approved limit is €12,450.75, not €12,540.75."}]}
{"case_id": "LANG-es", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-es/seed-1731/checkpoint.json", "checkpoint_sha256": "da5a21adf0cddba415463bf55c147f85447cc0a789b192712b8085e87da8030c", "completed_utc": "2026-08-02T02:16:11.553662+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 792, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.144315784102076, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 792, "prompt_tokens_per_second": 310.4190063428291, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2218889216, "peak_system_memory_used_bytes": 40262434816, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 148.06, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 142.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 138.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 141.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 141.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 141.29, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 21, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5609573150286451}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "es", "max_score": 10.0, "passed": false, "score": 7.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 142, "path": "results/quants/UD-IQ2_M/cases/LANG-es/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dbb204af3a0902e8046eb17d2711627ef6edaa818702587394eeec1640a89069", "utf8": "Mara Chen no aprobó el envío VX-44. Permanece en espera hasta el 17 de agosto de 2026. El límite aprobado es €12,450.75, no €12,540.75."}]}
{"case_id": "LANG-es", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-es/seed-2731/checkpoint.json", "checkpoint_sha256": "41cb823c8e011eb1f7d303705b61b13ffb9d1830ea5b878d77bb3ff174f6096b", "completed_utc": "2026-08-02T02:16:19.132565+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 236, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.44428053967822, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 236, "prompt_tokens_per_second": 309.9765049655472, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2261417984, "peak_system_memory_used_bytes": 40261947392, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 150.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 141.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 138.36, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 140.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 140.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 141.13, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5562567870365456}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "es", "max_score": 10.0, "passed": false, "score": 5.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 142, "path": "results/quants/UD-IQ2_M/cases/LANG-es/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ef22d3cbadcc1dae698c3aa4498d7738f3c1ffbb79ab3285738590af6538d425", "utf8": "Mara Chen no aprobó el envío VX-44. Permanece en espera hasta el 17 de agosto de 2026. El límite aprobado es €12.450,75, no €12.540,75."}]}
{"case_id": "LANG-es", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-es/seed-731/checkpoint.json", "checkpoint_sha256": "37762a60430a65083d804b25607ec03a9e1a0c0e17be513c4fa597cfafdc72df", "completed_utc": "2026-08-02T02:15:48.125092+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 341, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.44347794272334, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 341, "prompt_tokens_per_second": 311.12212038642554, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2179309568, "peak_system_memory_used_bytes": 40170655744, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 150.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 141.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 138.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 155.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 141.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 138.85, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 10, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5553602969739586}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "es", "max_score": 10.0, "passed": false, "score": 5.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 144, "path": "results/quants/UD-IQ2_M/cases/LANG-es/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fd378cacd5bb8dc8bd0c4873f72df01cae314e251f4a41224aca10fef02ad0a9", "utf8": "Mara Chen no aprobó el envío VX-44. Permanece en espera hasta el 17 de agosto de 2026. El límite aprobado es 12.450,75 €, no 12.540,75 €."}]}
{"case_id": "LANG-fr", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-fr/seed-1731/checkpoint.json", "checkpoint_sha256": "c9820108b57845c2039c78a44eb9b15060a3ec51e46db325494bd40392fd3e8a", "completed_utc": "2026-08-02T02:17:02.352573+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 426, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.33109088724082, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 426, "prompt_tokens_per_second": 313.2726483236939, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2344386560, "peak_system_memory_used_bytes": 40420913152, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 145.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 141.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 140.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 144.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 140.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 156.64, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 12, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5470853869337589}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "fr", "max_score": 10.0, "passed": false, "score": 7.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 154, "path": "results/quants/UD-IQ2_M/cases/LANG-fr/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "91934ccb735165c181b28deb3c52343db6389f4a337984262e66ed99602d34da", "utf8": "Mara Chen n'a pas approuvé l'expédition VX-44. Elle reste en attente jusqu'au 17 août 2026. La limite approuvée est de €12,450.75, pas €12,540.75."}]}
{"case_id": "LANG-fr", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-fr/seed-2731/checkpoint.json", "checkpoint_sha256": "082a9c5a5735e085a75bf343710f5c404fcb9c8f26bc2cb5545b616abff72e0b", "completed_utc": "2026-08-02T02:17:09.423370+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 221, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.611537106738254, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 221, "prompt_tokens_per_second": 312.76414218495444, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2384371712, "peak_system_memory_used_bytes": 40493604864, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 146.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 142.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 138.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 141.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 139.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 141.41, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5443560790736228}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "fr", "max_score": 10.0, "passed": false, "score": 5.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 159, "path": "results/quants/UD-IQ2_M/cases/LANG-fr/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "87593e90c1fe3ab911f6bf43777ad3f9b2090268d8276c0477f6244bc0fd10c0", "utf8": "Mara Chen n'a pas approuvé l'expédition VX-44. Elle reste en attente jusqu'au 17 août 2026. La limite approuvée est de 12 450,75 €, et non 12 540,75 €."}]}
{"case_id": "LANG-fr", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-fr/seed-731/checkpoint.json", "checkpoint_sha256": "8889a1b67ec3b2497fa626019b54cf90a6d109176695c7a4216d93130c47281d", "completed_utc": "2026-08-02T02:16:49.373788+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 1024, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.89259147140559, "finish_reason": "length", "output_token_cap": 1024, "output_tokens": 1024, "prompt_tokens_per_second": 311.9588846087779, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2300235776, "peak_system_memory_used_bytes": 40484212736, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 156.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 144.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 140.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 143.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 146.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 146.6, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 27, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5491014450090006}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 0.0, "hold_status": 0.0, "name": 0.0, "negation": 0.0, "semantic_adequacy": 0.0, "shipment_identifier": 0.0}, "language": "fr", "max_score": 10.0, "passed": false, "score": 0.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/LANG-fr/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "LANG-ja", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-ja/seed-1731/checkpoint.json", "checkpoint_sha256": "4d437fc023ce21f142153cb2342ed460935d03dde71ebf31b48e1007651b9b41", "completed_utc": "2026-08-02T02:17:45.749362+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 187, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.6735568138592, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 187, "prompt_tokens_per_second": 311.9767478596026, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2467143680, "peak_system_memory_used_bytes": 40583901184, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 147.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 145.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 141.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 144.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 144.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 144.06, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5583288369234651}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 0.0, "negation": 0.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "ja", "max_score": 10.0, "passed": false, "score": 5.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 192, "path": "results/quants/UD-IQ2_M/cases/LANG-ja/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "048d6b60c99cea2798c24f8eec3c475738c7792ade931c55f3ca04ebb8d32dce", "utf8": "マラ・チェンは出荷VX-44を承認しなかった。それは2026年8月17日まで保留のままである。承認された上限は€12,450.75であり、€12,540.75ではない。"}]}
{"case_id": "LANG-ja", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-ja/seed-2731/checkpoint.json", "checkpoint_sha256": "b5364c5c577c59b52cb4a3a0be3d27b30c38aaacbf130a7e56cdee30b2d470d6", "completed_utc": "2026-08-02T02:17:54.430712+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 279, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.50490133073388, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 279, "prompt_tokens_per_second": 314.5029658823998, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2505527296, "peak_system_memory_used_bytes": 40614121472, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 151.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 147.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 137.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 142.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 153.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 154.95, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5427697210106999}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 0.0, "negation": 0.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "ja", "max_score": 10.0, "passed": false, "score": 5.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 195, "path": "results/quants/UD-IQ2_M/cases/LANG-ja/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "14397fb31cd04beb3972ab2d6df9600e0728c86b106bdfeb2dd69a93a0b82e59", "utf8": "マラ・チェンは出荷VX-44を承認しなかった。それは2026年8月17日まで保留のままである。承認された限度額は€12,450.75であり、€12,540.75ではない。"}]}
{"case_id": "LANG-ja", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-ja/seed-731/checkpoint.json", "checkpoint_sha256": "bb838436e742353596a1f1802b436cc846e5ad4cf0eafc58b4ce39b2281b320b", "completed_utc": "2026-08-02T02:17:39.617086+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 1024, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.886633431697284, "finish_reason": "length", "output_token_cap": 1024, "output_tokens": 1024, "prompt_tokens_per_second": 313.801506644449, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2422996992, "peak_system_memory_used_bytes": 40621506560, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 148.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 142.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 138.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 152.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 150.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 140.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 27, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5479731549276039}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 0.0, "hold_status": 0.0, "name": 0.0, "negation": 0.0, "semantic_adequacy": 0.0, "shipment_identifier": 0.0}, "language": "ja", "max_score": 10.0, "passed": false, "score": 0.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/LANG-ja/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "LANG-zh-Hans", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-zh-Hans/seed-1731/checkpoint.json", "checkpoint_sha256": "86e841ee549b8debab5b8f52fe42b87613750b4d503454685cee9331bfe30765", "completed_utc": "2026-08-02T02:18:11.888450+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 215, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.61458853057852, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 215, "prompt_tokens_per_second": 314.66269527165554, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2583941120, "peak_system_memory_used_bytes": 40570728448, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 145.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 143.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 142.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 142.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 139.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 141.26, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5598855320131406}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 0.0, "name": 0.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "zh-Hans", "max_score": 10.0, "passed": false, "score": 5.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 138, "path": "results/quants/UD-IQ2_M/cases/LANG-zh-Hans/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "286b9f5e233c3318eef079375c272c87a7d0cb9e3b4f5f43b495226c43e2399c", "utf8": "未批准发货 VX-44。该货件保持保留状态，直至 2026 年 8 月 17 日。批准限额为 €12,450.75，而非 €12,540.75。"}]}
{"case_id": "LANG-zh-Hans", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-zh-Hans/seed-2731/checkpoint.json", "checkpoint_sha256": "7828a6552e361079fd51775e4459a5252be2a32083de5431454277555a97a754", "completed_utc": "2026-08-02T02:18:18.732422+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 210, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.6031650874685, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 210, "prompt_tokens_per_second": 315.54768101853307, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2622427136, "peak_system_memory_used_bytes": 40595185664, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 145.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 140.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 136.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 141.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 139.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 140.69, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5566129820654169}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 0.0, "name": 1.0, "negation": 2.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "zh-Hans", "max_score": 10.0, "passed": false, "score": 7.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 154, "path": "results/quants/UD-IQ2_M/cases/LANG-zh-Hans/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "eb1160f874548ec721fe796075d95bafd24c46baafcffbedff7c884c3344017d", "utf8": "Mara Chen 未批准发运 VX-44。该发运将保持搁置状态，直至 2026 年 8 月 17 日。批准的限额是 €12,450.75，不是 €12,540.75。"}]}
{"case_id": "LANG-zh-Hans", "checkpoint_path": "results/quants/UD-IQ2_M/cases/LANG-zh-Hans/seed-731/checkpoint.json", "checkpoint_sha256": "cef53a64fc33b7fa9e67926919b97ed858656b26a1d31f816f55159a3febf0fa", "completed_utc": "2026-08-02T02:18:04.941726+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 340, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.50292027182289, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 340, "prompt_tokens_per_second": 316.0035172565399, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2544558080, "peak_system_memory_used_bytes": 40640753664, "per_gpu": {"0": {"peak_memory_used_mib": 17597.0, "peak_power_draw_w": 149.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 141.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16515.0, "peak_power_draw_w": 140.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19085.0, "peak_power_draw_w": 142.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16545.0, "peak_power_draw_w": 139.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15125.0, "peak_power_draw_w": 144.95, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5573754360666499}, "quant": "UD-IQ2_M", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 0.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "zh-Hans", "max_score": 10.0, "passed": false, "score": 6.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 141, "path": "results/quants/UD-IQ2_M/cases/LANG-zh-Hans/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b8d41ec0d8c38838b9dd850c936787c4709ee625ebd8392ab850276a9469cf4b", "utf8": "Mara Chen未批准发运VX-44。该发运仍处于搁置状态，直至2026年8月17日。批准限额为€12,450.75，而非€12,540.75。"}]}
{"case_id": "PY-PATCH-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/PY-PATCH-01/seed-1731/checkpoint.json", "checkpoint_sha256": "8757276047fc1598aa122831765633982ab5b0619713db890cc5726f8f9c7ad3", "completed_utc": "2026-08-02T02:22:31.977683+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.76015040588155, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 636.2211332585164, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2749386752, "peak_system_memory_used_bytes": 47055970304, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 159.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 153.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 148.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19111.0, "peak_power_draw_w": 144.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 164.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15151.0, "peak_power_draw_w": 166.55, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 114, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.38933791895397}, "quant": "UD-IQ2_M", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "python_patch"}, "seed": 1731, "status": "completed", "task": "python_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/PY-PATCH-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "PY-PATCH-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/PY-PATCH-01/seed-2731/checkpoint.json", "checkpoint_sha256": "a2f78f1de0d51456336dfc0fcb3cc9d10190751cc5f4c5051663ae33bf40da92", "completed_utc": "2026-08-02T02:24:38.574659+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.80328925372653, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 636.0248072309155, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2819506176, "peak_system_memory_used_bytes": 48407142400, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 152.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 150.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 150.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19111.0, "peak_power_draw_w": 152.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 159.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15151.0, "peak_power_draw_w": 143.17, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 114, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.390637801028788}, "quant": "UD-IQ2_M", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "python_patch"}, "seed": 2731, "status": "completed", "task": "python_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/PY-PATCH-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "PY-PATCH-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/PY-PATCH-01/seed-731/checkpoint.json", "checkpoint_sha256": "199e5dd9da2ce2e04dafc8c988990895424c473a016c2eebdbe4a7d6986bdd09", "completed_utc": "2026-08-02T02:20:25.319339+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.810328473522375, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 636.4516949773165, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2679271424, "peak_system_memory_used_bytes": 46381465600, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 181.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 145.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 152.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19111.0, "peak_power_draw_w": 156.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 146.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15151.0, "peak_power_draw_w": 154.15, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 114, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3804598280694336}, "quant": "UD-IQ2_M", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "python_patch"}, "seed": 731, "status": "completed", "task": "python_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/PY-PATCH-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-CF-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-01/seed-1731/checkpoint.json", "checkpoint_sha256": "b52599738d9a32a14f5e887abc19e68da6b292e88db408db530bbb66e03316b2", "completed_utc": "2026-08-02T02:25:08.094786+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 523, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.17145546543907, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 523, "prompt_tokens_per_second": 327.76243778509337, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2930348032, "peak_system_memory_used_bytes": 48792162304, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 147.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 142.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 137.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 139.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.52, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6763088010484353}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ2_M/cases/R-CF-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "76061ca5fbef0891168215924b8e409ae5f5476fc3e95cfa5873ba807226fc13", "utf8": "[\"P3\", \"P4\", \"P5\", \"P6\", \"P7\", \"P9\"]"}]}
{"case_id": "R-CF-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-01/seed-2731/checkpoint.json", "checkpoint_sha256": "2ba3d3c5626838384d2245def586b9fe9044483a3857e71ca28d70a69521305c", "completed_utc": "2026-08-02T02:25:32.232630+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 809, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.036566655530606, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 809, "prompt_tokens_per_second": 328.1927147621083, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2971308032, "peak_system_memory_used_bytes": 45941985280, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 152.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 147.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 148.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 150.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 145.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 149.51, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 22, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6854646250139922}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 31, "path": "results/quants/UD-IQ2_M/cases/R-CF-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a6100b4182517cae95ba792e4147ca4d36ff0eb0a03438e07ad7de1c276bde52", "utf8": "[\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P9\"]"}]}
{"case_id": "R-CF-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-01/seed-731/checkpoint.json", "checkpoint_sha256": "b6aa3e90fb7bb0130cda610a531344860d4e023c9c3d85c9c387db896f06e62c", "completed_utc": "2026-08-02T02:24:52.163703+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 440, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.19023682069646, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 440, "prompt_tokens_per_second": 326.92559173532106, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2889932800, "peak_system_memory_used_bytes": 47027027968, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 147.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 141.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 143.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 146.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 159.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.85, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7054364990908653}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 31, "path": "results/quants/UD-IQ2_M/cases/R-CF-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a6100b4182517cae95ba792e4147ca4d36ff0eb0a03438e07ad7de1c276bde52", "utf8": "[\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P9\"]"}]}
{"case_id": "R-CF-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-02/seed-1731/checkpoint.json", "checkpoint_sha256": "9743be213dd6107d35e3e5de8eb3b7fe5fd28c4861be96a92f830fb1182a5483", "completed_utc": "2026-08-02T02:26:07.429971+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 483, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.18406108127831, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 483, "prompt_tokens_per_second": 329.23899217535427, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3056291840, "peak_system_memory_used_bytes": 44120367104, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 149.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 151.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 142.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 148.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 145.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 147.22, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6782300290651619}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 42, "path": "results/quants/UD-IQ2_M/cases/R-CF-02/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "7e7579b7f34079f8e56bcdbe61a4c4bfc9ed2db6092ebfe5f52dda843aaeb14a", "utf8": "[\"P1\", \"P3\", \"P5\", \"P6\", \"P7\", \"P8\", \"P9\"]"}]}
{"case_id": "R-CF-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-02/seed-2731/checkpoint.json", "checkpoint_sha256": "8ce302e11c17dcd94ef8b14bc74a1027d54fdb4f758f43fe90b3c95232b86189", "completed_utc": "2026-08-02T02:26:22.038470+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 478, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.16623476448628, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 478, "prompt_tokens_per_second": 328.8803469126157, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3096969216, "peak_system_memory_used_bytes": 44609429504, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 147.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 143.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 140.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 140.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 144.37, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6764452180359513}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ2_M/cases/R-CF-02/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "155ba39cf2c1ba52f3d26c7f08cad012d6a2fe8e0df297ae8c0ceaff5469ce73", "utf8": "[\"P1\",\"P3\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-02/seed-731/checkpoint.json", "checkpoint_sha256": "ad9158b5cb68d3709c7b62538d46d82b0bbf9ac967a0483901e9483d1ff72d06", "completed_utc": "2026-08-02T02:25:52.645499+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 682, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.13362600708453, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 682, "prompt_tokens_per_second": 329.39770032200636, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3014246400, "peak_system_memory_used_bytes": 44472193024, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 150.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 141.06, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 138.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.36, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 141.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 142.75, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 18, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6804815019713715}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ2_M/cases/R-CF-02/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "155ba39cf2c1ba52f3d26c7f08cad012d6a2fe8e0df297ae8c0ceaff5469ce73", "utf8": "[\"P1\",\"P3\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-03/seed-1731/checkpoint.json", "checkpoint_sha256": "082c483156a9dcd08f5506292513ebf4c50b58e16d3dd805076179f5116a4220", "completed_utc": "2026-08-02T02:26:51.273774+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 463, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.17079866578854, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 463, "prompt_tokens_per_second": 327.114911257935, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3178377216, "peak_system_memory_used_bytes": 44339159040, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 153.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 141.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 142.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 142.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 145.4, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 12, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6577241950435564}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 30, "path": "results/quants/UD-IQ2_M/cases/R-CF-03/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1afe6295906b2c3400159454af0ffcd4cfbf8e48df321b0f2475a39ad3dde4a3", "utf8": "[\"P3\", \"P4\", \"P5\", \"P6\", \"P9\"]"}]}
{"case_id": "R-CF-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-03/seed-2731/checkpoint.json", "checkpoint_sha256": "fa12ccd338d189409565a5bef2335de7cf534c90c8ab9a198628326b959cbd90", "completed_utc": "2026-08-02T02:27:07.673210+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 540, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.22717909947979, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 540, "prompt_tokens_per_second": 325.27866075368036, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3218923520, "peak_system_memory_used_bytes": 44381540352, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 146.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 141.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 142.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 140.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 143.18, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 15, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6932671240065247}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 30, "path": "results/quants/UD-IQ2_M/cases/R-CF-03/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1afe6295906b2c3400159454af0ffcd4cfbf8e48df321b0f2475a39ad3dde4a3", "utf8": "[\"P3\", \"P4\", \"P5\", \"P6\", \"P9\"]"}]}
{"case_id": "R-CF-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-03/seed-731/checkpoint.json", "checkpoint_sha256": "1946f7a36db5d940367698d3b225e0f43635c6a58a73f42d6a10ac4a6db4f90f", "completed_utc": "2026-08-02T02:26:37.088701+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 495, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.18694110964258, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 495, "prompt_tokens_per_second": 327.44685472706817, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3137617920, "peak_system_memory_used_bytes": 44579184640, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 147.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 142.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 139.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 140.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 143.21, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6552942749112844}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 30, "path": "results/quants/UD-IQ2_M/cases/R-CF-03/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1afe6295906b2c3400159454af0ffcd4cfbf8e48df321b0f2475a39ad3dde4a3", "utf8": "[\"P3\", \"P4\", \"P5\", \"P6\", \"P9\"]"}]}
{"case_id": "R-CF-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-04/seed-1731/checkpoint.json", "checkpoint_sha256": "9e783c2c1b1daa3171ea93e6a7f2062feb05d583b407bba582801a42355f0dff", "completed_utc": "2026-08-02T02:27:36.239971+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 553, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.23863020498624, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 553, "prompt_tokens_per_second": 328.07663870280095, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3299942400, "peak_system_memory_used_bytes": 44365225984, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 145.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 141.06, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 137.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 139.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.26, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 15, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6767894310178235}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ2_M/cases/R-CF-04/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-04/seed-2731/checkpoint.json", "checkpoint_sha256": "688548903fd5f254faab94a8452030c227582ca758996f7dd0def9d3e77e24b2", "completed_utc": "2026-08-02T02:27:48.219479+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 386, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.28641507757435, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 386, "prompt_tokens_per_second": 329.4347542416733, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3341037568, "peak_system_memory_used_bytes": 44397154304, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 153.36, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 142.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 137.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 141.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 160.47, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 11, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6702386389952153}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ2_M/cases/R-CF-04/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-04/seed-731/checkpoint.json", "checkpoint_sha256": "9afc853007ceeed12df723c2bcc365f90c74376f9588784eff3f7f98dd0d02d8", "completed_utc": "2026-08-02T02:27:19.490221+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 378, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.273342274223545, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 378, "prompt_tokens_per_second": 328.558286239979, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3259977728, "peak_system_memory_used_bytes": 44420767744, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 149.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 143.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 146.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 148.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 145.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.84, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 11, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6806436190381646}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 54, "path": "results/quants/UD-IQ2_M/cases/R-CF-04/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "5d9103e860a76affa701ceb85a56b4c2c47728e53d1fa6957e9c0a40da52e971", "utf8": "[\"P1\", \"P2\", \"P3\", \"P4\", \"P5\", \"P6\", \"P7\", \"P8\", \"P9\"]"}]}
{"case_id": "R-CF-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-05/seed-1731/checkpoint.json", "checkpoint_sha256": "fda541fc5e7934fa8cef609bbff19237d0e7f3c29eb735dab47d440a329b107d", "completed_utc": "2026-08-02T02:28:35.868635+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 829, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.952479174288364, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 829, "prompt_tokens_per_second": 328.4361632247625, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3423490048, "peak_system_memory_used_bytes": 44505960448, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 154.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 149.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 144.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 150.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 143.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 143.26, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 22, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6792319359956309}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ2_M/cases/R-CF-05/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b7c5d63364adfb6a5a28ccff4ba1ab0a92bb532e4161877d32f5177838853194", "utf8": "[\"P1\",\"P3\",\"P4\",\"P5\",\"P6\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-05/seed-2731/checkpoint.json", "checkpoint_sha256": "4fee1ac8c896b2eaa19e8cecb294548ba6f8f1c9e1c65b8ffc524cb7dea86f5e", "completed_utc": "2026-08-02T02:28:58.828015+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 767, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.03071056074074, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 767, "prompt_tokens_per_second": 328.85133032394265, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3466469376, "peak_system_memory_used_bytes": 44553330688, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 147.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 145.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 140.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 148.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 151.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 147.13, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 20, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6850331490859389}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ2_M/cases/R-CF-05/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b7c5d63364adfb6a5a28ccff4ba1ab0a92bb532e4161877d32f5177838853194", "utf8": "[\"P1\",\"P3\",\"P4\",\"P5\",\"P6\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-05/seed-731/checkpoint.json", "checkpoint_sha256": "c70a07ad9421695fd39acec7d9c4eca9c018fb7d71c8171caa73d9c43ebe33f9", "completed_utc": "2026-08-02T02:28:11.128193+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 767, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.057882529491586, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 767, "prompt_tokens_per_second": 329.48240724668426, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3380948992, "peak_system_memory_used_bytes": 44505542656, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 150.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 142.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 139.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 140.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 155.28, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 20, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6729974129702896}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ2_M/cases/R-CF-05/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b7c5d63364adfb6a5a28ccff4ba1ab0a92bb532e4161877d32f5177838853194", "utf8": "[\"P1\",\"P3\",\"P4\",\"P5\",\"P6\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-06/seed-1731/checkpoint.json", "checkpoint_sha256": "1814d2dd0eaf89643f62c3a3b48ee6ca6dd476eec71f1a3eb151b094c7f1b7ec", "completed_utc": "2026-08-02T02:29:38.545601+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 530, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.21139392873187, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 530, "prompt_tokens_per_second": 332.0959565720672, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3551703040, "peak_system_memory_used_bytes": 44558237696, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 149.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 140.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 143.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 142.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.89, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6843099870020524}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ2_M/cases/R-CF-06/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-06/seed-2731/checkpoint.json", "checkpoint_sha256": "2aff6bae47a524d9dd599c4739e004172f0a440ec74bb6155e8f1decf08e675d", "completed_utc": "2026-08-02T02:30:00.327342+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 725, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.055425044611056, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 725, "prompt_tokens_per_second": 332.20150576720977, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3592630272, "peak_system_memory_used_bytes": 44763394048, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 146.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 148.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 146.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 150.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 146.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.74, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 19, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6856437349924818}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ2_M/cases/R-CF-06/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-06/seed-731/checkpoint.json", "checkpoint_sha256": "608074852d7fbe00abb3361b244f96502d665210858e73ba5b241db366945721", "completed_utc": "2026-08-02T02:29:22.431478+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 790, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.031865916464426, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 790, "prompt_tokens_per_second": 331.9078317482607, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3508998144, "peak_system_memory_used_bytes": 44524236800, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 152.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 144.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 142.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 148.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 143.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 146.24, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 21, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6895342109492049}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ2_M/cases/R-CF-06/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-07/seed-1731/checkpoint.json", "checkpoint_sha256": "439fcab94aa3d566200a1c7200b2b532dde2c7d85c00b6c5194bfc44ded00d85", "completed_utc": "2026-08-02T02:30:39.615598+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 772, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.0423022320494, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 772, "prompt_tokens_per_second": 339.57204401742285, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3675910144, "peak_system_memory_used_bytes": 44920590336, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 154.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 150.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 146.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 149.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 147.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 148.0, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 21, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6708375669550151}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 30, "path": "results/quants/UD-IQ2_M/cases/R-CF-07/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "909bfa8687c5ddbc479433905568f34e553d34b1de20fb6cab8dd83bfc413fd7", "utf8": "[\"P1\", \"P2\", \"P4\", \"P5\", \"P8\"]"}]}
{"case_id": "R-CF-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-07/seed-2731/checkpoint.json", "checkpoint_sha256": "a9984ee19c5d58b3bc8310ad4d9a2c49a05fd651670280ba8b79b2dbef39920b", "completed_utc": "2026-08-02T02:30:58.276597+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 619, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.12398112792466, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 619, "prompt_tokens_per_second": 338.019804169648, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3718565888, "peak_system_memory_used_bytes": 44685037568, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 146.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 149.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 145.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 148.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 145.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.46, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 17, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6931842470075935}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 30, "path": "results/quants/UD-IQ2_M/cases/R-CF-07/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "909bfa8687c5ddbc479433905568f34e553d34b1de20fb6cab8dd83bfc413fd7", "utf8": "[\"P1\", \"P2\", \"P4\", \"P5\", \"P8\"]"}]}
{"case_id": "R-CF-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-CF-07/seed-731/checkpoint.json", "checkpoint_sha256": "20f6b53fe26a00b23058af522408869cc480b726e1e3f4b07d1e8368413eb113", "completed_utc": "2026-08-02T02:30:16.557988+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 535, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.174437927274916, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 535, "prompt_tokens_per_second": 338.29552801766727, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3634905088, "peak_system_memory_used_bytes": 45970124800, "per_gpu": {"0": {"peak_memory_used_mib": 17890.0, "peak_power_draw_w": 147.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16838.0, "peak_power_draw_w": 141.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16808.0, "peak_power_draw_w": 138.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19380.0, "peak_power_draw_w": 144.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16838.0, "peak_power_draw_w": 140.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15420.0, "peak_power_draw_w": 140.58, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6910126250004396}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 30, "path": "results/quants/UD-IQ2_M/cases/R-CF-07/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "909bfa8687c5ddbc479433905568f34e553d34b1de20fb6cab8dd83bfc413fd7", "utf8": "[\"P1\", \"P2\", \"P4\", \"P5\", \"P8\"]"}]}
{"case_id": "R-GRAPH-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-01/seed-1731/checkpoint.json", "checkpoint_sha256": "3eda412dbf955f10e9a2c778b29050aec4b8ea4b6667168ae3ffd18a4f327293", "completed_utc": "2026-08-02T02:31:13.092591+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 221, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.32924377354051, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 221, "prompt_tokens_per_second": 327.55859659339063, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3798904832, "peak_system_memory_used_bytes": 44773314560, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 146.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 141.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 145.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 150.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 151.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 151.17, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6785970509517938}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "80bed85c4ee03d4fc0818522004a326c9facf394a916a487a3744b49804b99a2", "utf8": "{\"reachable\":[\"E\",\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-01/seed-2731/checkpoint.json", "checkpoint_sha256": "3df3754d7c619b527f26e16d7e0f0fc3b36834352a83700b18944dd0f1d481b2", "completed_utc": "2026-08-02T02:31:18.617098+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 158, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.46379460806671, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 158, "prompt_tokens_per_second": 325.9339503509695, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3837661184, "peak_system_memory_used_bytes": 44726435840, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 144.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 142.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 138.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 142.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 138.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.65, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6892673049587756}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "80bed85c4ee03d4fc0818522004a326c9facf394a916a487a3744b49804b99a2", "utf8": "{\"reachable\":[\"E\",\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-01/seed-731/checkpoint.json", "checkpoint_sha256": "7a501052f474a5cdc7f0c28c4b3a6af1be67f3959b5fcdff7ed37c3de1583470", "completed_utc": "2026-08-02T02:31:05.840062+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 229, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.287451735934596, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 229, "prompt_tokens_per_second": 325.94472796212114, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3760111616, "peak_system_memory_used_bytes": 44951875584, "per_gpu": {"0": {"peak_memory_used_mib": 17623.0, "peak_power_draw_w": 151.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 148.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16541.0, "peak_power_draw_w": 137.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 147.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16571.0, "peak_power_draw_w": 148.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 152.79, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6895934559870511}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "80bed85c4ee03d4fc0818522004a326c9facf394a916a487a3744b49804b99a2", "utf8": "{\"reachable\":[\"E\",\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-02/seed-1731/checkpoint.json", "checkpoint_sha256": "3b64d2d5a78c13934d8349d02f1c2a76befd485119b5e27ec7680f4f58a9d002", "completed_utc": "2026-08-02T02:31:28.358723+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 123, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.47427812007091, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 123, "prompt_tokens_per_second": 327.6686120918296, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3876114432, "peak_system_memory_used_bytes": 44901683200, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 149.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 136.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 138.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.6, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5990936019225046}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 44, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-02/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "027d14d5a4eca73d6cd0cc389f7448a0a3c7d3ac648fe299e67b153bf55832af", "utf8": "{\"reachable\":[\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-02/seed-2731/checkpoint.json", "checkpoint_sha256": "96a0a8afeab39c9f65c113ed0c154db9907bbb1835520ea9851f78cce20838d4", "completed_utc": "2026-08-02T02:31:34.036123+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 168, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.54806224973529, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 168, "prompt_tokens_per_second": 328.8561830988583, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3876241408, "peak_system_memory_used_bytes": 44897767424, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 144.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 147.08, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5891448069596663}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 44, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-02/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "027d14d5a4eca73d6cd0cc389f7448a0a3c7d3ac648fe299e67b153bf55832af", "utf8": "{\"reachable\":[\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-02/seed-731/checkpoint.json", "checkpoint_sha256": "159d6ab16ea43fe0b2b1cdee58292c545e56c3f1fc63ee5f9d85588c16509538", "completed_utc": "2026-08-02T02:31:23.979808+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 153, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.5227332723333, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 153, "prompt_tokens_per_second": 330.433819814957, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3875991552, "peak_system_memory_used_bytes": 44840779776, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 144.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 139.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 142.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 137.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.75, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6269733220105991}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 44, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-02/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "027d14d5a4eca73d6cd0cc389f7448a0a3c7d3ac648fe299e67b153bf55832af", "utf8": "{\"reachable\":[\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-03/seed-1731/checkpoint.json", "checkpoint_sha256": "87c8e12f25b9a3e251d5555f34f85ee19f09691a0b147c3306e2f30c7e08ffb8", "completed_utc": "2026-08-02T02:31:49.588966+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 188, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.28673569728185, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 188, "prompt_tokens_per_second": 322.0223638979617, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3953684480, "peak_system_memory_used_bytes": 44926898176, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 148.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 139.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 145.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 151.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 138.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 143.54, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6816495229722932}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-03/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b22e387e29cad42a2cecb3c833850242fe106aa67d7430530bac5d1e558d4d68", "utf8": "{\"reachable\":[\"E\",\"F\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-03/seed-2731/checkpoint.json", "checkpoint_sha256": "5e0450fe94ba1f81350bb640ff0654fc60255675c0c2644603832ca426a88f92", "completed_utc": "2026-08-02T02:31:55.974850+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 191, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.31636249601998, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 191, "prompt_tokens_per_second": 320.41113641780345, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3953827840, "peak_system_memory_used_bytes": 44974813184, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 157.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 139.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 142.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 154.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 158.91, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6553818769752979}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-03/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b22e387e29cad42a2cecb3c833850242fe106aa67d7430530bac5d1e558d4d68", "utf8": "{\"reachable\":[\"E\",\"F\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-03/seed-731/checkpoint.json", "checkpoint_sha256": "bb417dc3e11b3173761972d73e7e6c166f0177c4977888e1053927d357fd14ed", "completed_utc": "2026-08-02T02:31:43.178985+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 286, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.39772187193055, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 286, "prompt_tokens_per_second": 322.02747227470377, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3914510336, "peak_system_memory_used_bytes": 44852039680, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 145.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 137.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 142.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 139.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.44, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6805343530140817}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-03/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b22e387e29cad42a2cecb3c833850242fe106aa67d7430530bac5d1e558d4d68", "utf8": "{\"reachable\":[\"E\",\"F\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-04/seed-1731/checkpoint.json", "checkpoint_sha256": "a110a90be013d9694762082abbc69db2253e3acfbef586e157d67c15bb215718", "completed_utc": "2026-08-02T02:32:09.115812+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 216, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.27433635255524, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 216, "prompt_tokens_per_second": 319.2095153308791, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3992584192, "peak_system_memory_used_bytes": 45308977152, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 155.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 149.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 151.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 150.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 152.77, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6391926440410316}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 47, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-04/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8ae9db64671b10d5aca89092eda3e058a63c807d44ddf14e23414eee8a384764", "utf8": "{\"reachable\":[\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-04/seed-2731/checkpoint.json", "checkpoint_sha256": "e88c25fdf7af63971623025f8cfa26fa66a00851a077dc1c109f99a67ec9eba4", "completed_utc": "2026-08-02T02:32:15.581595+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 192, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.27598034613454, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 192, "prompt_tokens_per_second": 318.5493359286834, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4031246336, "peak_system_memory_used_bytes": 45390561280, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 145.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 141.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 139.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.45, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6767036040546373}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 47, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-04/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8ae9db64671b10d5aca89092eda3e058a63c807d44ddf14e23414eee8a384764", "utf8": "{\"reachable\":[\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-04/seed-731/checkpoint.json", "checkpoint_sha256": "6b694ff6101344616e79d927e2f608e76502dc7ceaad004cb703e955e0a90e2f", "completed_utc": "2026-08-02T02:32:01.997680+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 178, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.530232834805574, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 178, "prompt_tokens_per_second": 320.192050870311, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3992440832, "peak_system_memory_used_bytes": 44957990912, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 145.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 136.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.26, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6605286099947989}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 47, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-04/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8ae9db64671b10d5aca89092eda3e058a63c807d44ddf14e23414eee8a384764", "utf8": "{\"reachable\":[\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-05/seed-1731/checkpoint.json", "checkpoint_sha256": "813876e727fac192ef6e5f1e3bdf51ed03ca7deb4d3057ae9e1fe99461e3e58b", "completed_utc": "2026-08-02T02:32:31.621837+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 222, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.24480293948007, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 222, "prompt_tokens_per_second": 346.7273974213868, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4109004800, "peak_system_memory_used_bytes": 45602816000, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 166.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 152.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 145.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 145.68, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7050685629947111}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-05/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fe0eb6dafe2fb00aee4831bc12e5fea8af950087540db3bed2ad0b4d1bd8eab1", "utf8": "{\"reachable\":[\"E\",\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-05/seed-2731/checkpoint.json", "checkpoint_sha256": "010fd186b6933643246f7849131d42a6269b64356e37b9e8c10bb0109df6f7c4", "completed_utc": "2026-08-02T02:32:46.375347+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 482, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.16442725683763, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 482, "prompt_tokens_per_second": 346.6278606342221, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4109172736, "peak_system_memory_used_bytes": 45514096640, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 146.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 146.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 139.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.6, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6703923729946837}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-05/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fe0eb6dafe2fb00aee4831bc12e5fea8af950087540db3bed2ad0b4d1bd8eab1", "utf8": "{\"reachable\":[\"E\",\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-05/seed-731/checkpoint.json", "checkpoint_sha256": "1c8defd9e734583f693bee73c0b65138662ec7d606ab76275cf0559d9f9ecacd", "completed_utc": "2026-08-02T02:32:24.362364+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 273, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.30555600819089, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 273, "prompt_tokens_per_second": 346.05557622064623, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4069761024, "peak_system_memory_used_bytes": 45459234816, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 146.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 139.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.29, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7108710060128942}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-05/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fe0eb6dafe2fb00aee4831bc12e5fea8af950087540db3bed2ad0b4d1bd8eab1", "utf8": "{\"reachable\":[\"E\",\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-06/seed-1731/checkpoint.json", "checkpoint_sha256": "b12c025715eaac355562b9b1c293ff5566fb34f0f73dddf1176b6b6ffa873012", "completed_utc": "2026-08-02T02:33:01.383738+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 182, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.24977857522332, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 182, "prompt_tokens_per_second": 329.89058889378776, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4189044736, "peak_system_memory_used_bytes": 45613764608, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 147.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 140.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 155.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 144.08, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6912346020108089}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 56, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-06/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ab97d253da1946994b0f7c0ae1d8bc073294787350658f34ac8d55d3fe98ee24", "utf8": "{\"reachable\":[\"D\",\"E\",\"F\",\"G\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-06/seed-2731/checkpoint.json", "checkpoint_sha256": "c2763c90f1d975c71930faba514916aac5b7da1fb4cfb080c0448d291075f569", "completed_utc": "2026-08-02T02:33:08.712667+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 224, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.304189724715584, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 224, "prompt_tokens_per_second": 331.176240694497, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4189175808, "peak_system_memory_used_bytes": 45531082752, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 152.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 137.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 144.18, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6530336959986016}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 56, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-06/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ab97d253da1946994b0f7c0ae1d8bc073294787350658f34ac8d55d3fe98ee24", "utf8": "{\"reachable\":[\"D\",\"E\",\"F\",\"G\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-06/seed-731/checkpoint.json", "checkpoint_sha256": "8230478dcd7823c420c09948028bc7eb78a375f399415d6ee7e15805d20b85a4", "completed_utc": "2026-08-02T02:32:55.262359+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 278, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.33193911976447, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 278, "prompt_tokens_per_second": 329.4805450326514, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4149870592, "peak_system_memory_used_bytes": 45544927232, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 147.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 146.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 149.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 147.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7348158529493958}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 56, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-06/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ab97d253da1946994b0f7c0ae1d8bc073294787350658f34ac8d55d3fe98ee24", "utf8": "{\"reachable\":[\"D\",\"E\",\"F\",\"G\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-07/seed-1731/checkpoint.json", "checkpoint_sha256": "b0a3d44539128d856f5886b6d0f998c1950a97cf2914adb2517e3d594450d20b", "completed_utc": "2026-08-02T02:33:22.934676+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 292, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.35743766102316, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 292, "prompt_tokens_per_second": 323.70354385010177, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4228141056, "peak_system_memory_used_bytes": 45690232832, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 144.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 141.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 146.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 147.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.41, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6557976399781182}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 41, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-07/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a7add69bd5722467ba5ca5f5212bc9d53463eeb8315d3b18cc337421f5c0dfa0", "utf8": "{\"reachable\":[],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-07/seed-2731/checkpoint.json", "checkpoint_sha256": "2a0c3a5bd8d84b0dfcba50053df0b4f997a15082bbf1906fa354ab120cb5ced9", "completed_utc": "2026-08-02T02:33:28.795299+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 172, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.48365668159319, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 172, "prompt_tokens_per_second": 324.61982627350557, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4267364352, "peak_system_memory_used_bytes": 45979787264, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 149.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 145.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 137.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 148.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 152.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 146.52, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6963522019796073}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 41, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-07/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a7add69bd5722467ba5ca5f5212bc9d53463eeb8315d3b18cc337421f5c0dfa0", "utf8": "{\"reachable\":[],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-07/seed-731/checkpoint.json", "checkpoint_sha256": "b0223874961a2890511c12f3f7e4cef8760422f1ff09d78098cdb31bc77e7050", "completed_utc": "2026-08-02T02:33:13.717288+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 141, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.489554492826585, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 141, "prompt_tokens_per_second": 325.8734036926022, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4227977216, "peak_system_memory_used_bytes": 45621489664, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 144.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 139.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 140.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.91, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6828315859893337}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 41, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-07/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a7add69bd5722467ba5ca5f5212bc9d53463eeb8315d3b18cc337421f5c0dfa0", "utf8": "{\"reachable\":[],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-08", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-08/seed-1731/checkpoint.json", "checkpoint_sha256": "fd943efb1346be985565c8723ac9b1ad3b4ed1e6ee6b0ce2f4c7a7bc11032ea2", "completed_utc": "2026-08-02T02:33:43.425542+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 241, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.32572887165958, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 241, "prompt_tokens_per_second": 329.2764189706913, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4305887232, "peak_system_memory_used_bytes": 46739836928, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 144.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 136.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 141.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 138.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.39, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.656692482996732}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-08/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d5ff5eb60c7028268031852d6b3bd3006c4ec539197f97efd2d9a6c29247110a", "utf8": "{\"reachable\":[\"D\",\"H\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-08", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-08/seed-2731/checkpoint.json", "checkpoint_sha256": "68e49f15774083d84edb73f0869847076ec160224e73775ef2776dfa9eb9f5f7", "completed_utc": "2026-08-02T02:33:51.317817+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 242, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.30453222661892, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 242, "prompt_tokens_per_second": 330.2225637478227, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4344791040, "peak_system_memory_used_bytes": 46808383488, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 149.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 137.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.65, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6904091129545122}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-08/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d5ff5eb60c7028268031852d6b3bd3006c4ec539197f97efd2d9a6c29247110a", "utf8": "{\"reachable\":[\"D\",\"H\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-08", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-GRAPH-08/seed-731/checkpoint.json", "checkpoint_sha256": "5c61d54b112a074cbe220c7b4a57c24db528fd8320f50e09cec492b1277846df", "completed_utc": "2026-08-02T02:33:35.627337+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 204, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.25254647806324, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 204, "prompt_tokens_per_second": 331.53372845030765, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4305760256, "peak_system_memory_used_bytes": 45963157504, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 144.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 137.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 142.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 139.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.05, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6796255450462922}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ2_M/cases/R-GRAPH-08/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d5ff5eb60c7028268031852d6b3bd3006c4ec539197f97efd2d9a6c29247110a", "utf8": "{\"reachable\":[\"D\",\"H\"],\"target_reachable\":false}"}]}
{"case_id": "R-INV-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-01/seed-1731/checkpoint.json", "checkpoint_sha256": "66968240251f49a623311787590df8eff1c15929cd8ae645d88826ee36558a7e", "completed_utc": "2026-08-02T02:35:53.498748+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.17207250332673, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 339.3869167053624, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4435419136, "peak_system_memory_used_bytes": 49112440832, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 170.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 145.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 143.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 149.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 144.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 144.53, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7600864510750398}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-01/seed-2731/checkpoint.json", "checkpoint_sha256": "53ce6348dce2a158dd119fafadab0421e956fa389045f72c2c73f140ccf65326", "completed_utc": "2026-08-02T02:36:54.573898+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.16812730683868, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 336.59838460942615, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4487041024, "peak_system_memory_used_bytes": 44049006592, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 151.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 143.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 148.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 146.53, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7808664849726483}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-01/seed-731/checkpoint.json", "checkpoint_sha256": "82225c786fd29d2d13aecdd707b2d2a70a4d40254d5e3d1b641626c640b54103", "completed_utc": "2026-08-02T02:34:52.409574+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.149929060525785, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 337.54487652071737, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4383793152, "peak_system_memory_used_bytes": 52920016896, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 161.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 148.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 143.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 167.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 144.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 145.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.76155124604702}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-02/seed-1731/checkpoint.json", "checkpoint_sha256": "5d12662c4af880784956a8b2b310c69bae359167811e7fdeb669b9204e7d2e50", "completed_utc": "2026-08-02T02:38:56.862921+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.1336797901832, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 336.1259445631277, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4590301184, "peak_system_memory_used_bytes": 47770234880, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 150.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 141.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 147.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 152.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 153.75, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7778912590583786}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-02/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-02/seed-2731/checkpoint.json", "checkpoint_sha256": "9b00206bc766fdd22ceae38daf4cb072e81d28e267fcfe330a7a939833eea656", "completed_utc": "2026-08-02T02:39:58.012076+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.15857025434675, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 337.7251736670454, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4641927168, "peak_system_memory_used_bytes": 47951630336, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 150.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 149.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 152.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 145.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 150.38, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7640615419950336}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-02/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-02/seed-731/checkpoint.json", "checkpoint_sha256": "4915ef087528c63729ade1bb2ee76d6f321b52adf0382bc364fb69b5400a1749", "completed_utc": "2026-08-02T02:37:55.712654+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.13254373382163, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 336.59222204383303, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4538675200, "peak_system_memory_used_bytes": 47568121856, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 149.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 144.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 141.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 147.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 144.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 154.55, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7801368670770898}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-02/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-03/seed-1731/checkpoint.json", "checkpoint_sha256": "f0cb1061b9b233843181376cc7d9afc750dbfaddc804eb3c56b096b06af8af63", "completed_utc": "2026-08-02T02:41:48.136257+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 1649, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.437059538731354, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 1649, "prompt_tokens_per_second": 337.83162980438277, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4745166848, "peak_system_memory_used_bytes": 43518795776, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 147.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 149.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 144.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 146.74, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 44, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7728015980683267}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 39, "path": "results/quants/UD-IQ2_M/cases/R-INV-03/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "bfbd7e99fe73dd530546ef00bf5040973799a5947a95012cee977dd00e0274ed", "utf8": "{\"A\": [1, 6], \"B\": [5, 3], \"C\": [3, 3]}"}]}
{"case_id": "R-INV-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-03/seed-2731/checkpoint.json", "checkpoint_sha256": "a72a6a6ceb1946e333931ef82a5118a984464418c4dce0a4d8adb5d79eb19802", "completed_utc": "2026-08-02T02:42:49.136753+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.201753370922646, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 338.42480951214435, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4794044416, "peak_system_memory_used_bytes": 42924179456, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 158.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 149.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 140.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 150.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 149.58, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7752123849932104}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-03/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-03/seed-731/checkpoint.json", "checkpoint_sha256": "c2e5066a043109b4768b82a77a94e14144b3c0c8243cc82bd09b80e7ac88d65a", "completed_utc": "2026-08-02T02:40:59.095373+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.19387073197507, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 338.7509868425715, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4693549056, "peak_system_memory_used_bytes": 48238833664, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 146.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 149.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 145.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 150.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 145.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 156.1, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 54, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7774851359426975}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-03/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-04/seed-1731/checkpoint.json", "checkpoint_sha256": "aa4bae15258f4b7d7191d4d4b3ae5b2bf90712f217614bcb995fa862dd8dea1b", "completed_utc": "2026-08-02T02:44:51.282004+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.18060111694127, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 337.9940688404823, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4897288192, "peak_system_memory_used_bytes": 43259260928, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 159.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 146.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 157.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 151.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 151.02, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7747753970324993}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-04/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-04/seed-2731/checkpoint.json", "checkpoint_sha256": "4c139704a152bbe7c15cac1b17cc218bb1f7de78a56d5cfa264a27f1a7a8b2bc", "completed_utc": "2026-08-02T02:45:52.355685+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.17538899744377, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 337.3866790139839, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4948910080, "peak_system_memory_used_bytes": 44293484544, "per_gpu": {"0": {"peak_memory_used_mib": 17894.0, "peak_power_draw_w": 154.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16840.0, "peak_power_draw_w": 148.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16810.0, "peak_power_draw_w": 140.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19380.0, "peak_power_draw_w": 153.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16587.0, "peak_power_draw_w": 146.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 147.27, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7769069039495662}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-04/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-04/seed-731/checkpoint.json", "checkpoint_sha256": "c5bfe419479e8d82142ea3debd9a747a9d7980b14a4d36ab4b721741aa88a79b", "completed_utc": "2026-08-02T02:43:50.207101+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.18677295726103, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 337.26860410310536, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4845662208, "peak_system_memory_used_bytes": 43329949696, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 153.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 141.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 145.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 147.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 160.95, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7794061519671232}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-04/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-05/seed-1731/checkpoint.json", "checkpoint_sha256": "93e079d6bbf9f2522fc4a3bafe4b1fc1e0432e59e3fa3bedb32023dc6a498957", "completed_utc": "2026-08-02T02:47:54.913094+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.97243365924365, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 335.3284675629337, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5052162048, "peak_system_memory_used_bytes": 43400249344, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 154.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 149.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 144.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 153.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 142.17, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 54, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7703347430797294}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-05/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-05/seed-2731/checkpoint.json", "checkpoint_sha256": "d6a1c6b235be7ef3269e4e314536c334e59d82f6e3c74f2815c398458cb2d499", "completed_utc": "2026-08-02T02:48:56.010745+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.181083737630075, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 331.0542762793377, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5103812608, "peak_system_memory_used_bytes": 43578277888, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 148.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 140.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 151.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 142.97, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8034716169349849}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-05/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-05/seed-731/checkpoint.json", "checkpoint_sha256": "7ee6f3128938b7df464912a4fd8c72203216c361becdc8de57048d67679466fa", "completed_utc": "2026-08-02T02:46:53.449427+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.1930840380356, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 335.65247189086097, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5000536064, "peak_system_memory_used_bytes": 43318870016, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 151.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 148.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 143.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 151.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 145.62, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 54, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7813309609191492}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-05/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-06/seed-1731/checkpoint.json", "checkpoint_sha256": "14697bbf663568e0bee92fd9d3046b933d196d24cdd3074108b80f0d2fb9bdbd", "completed_utc": "2026-08-02T02:50:58.020179+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.2158659939618, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 336.8920789483329, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5207056384, "peak_system_memory_used_bytes": 45515034624, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 148.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 147.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 148.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 151.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 147.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 146.81, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7685720209265128}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-06/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-06/seed-2731/checkpoint.json", "checkpoint_sha256": "299239679a4d5ee6f0c23cc340a812669b1a79fe06ca1b7dd86ec6361c1d9536", "completed_utc": "2026-08-02T02:51:59.083723+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.208149406765045, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 335.02199370326866, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5258682368, "peak_system_memory_used_bytes": 48954470400, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 153.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 144.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 149.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 143.69, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7871073000133038}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-06/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-06/seed-731/checkpoint.json", "checkpoint_sha256": "42843e5ca5c97dff5ac1c98965e2385909a84cd276d5c19d0410670ad1fd62ca", "completed_utc": "2026-08-02T02:49:57.069021+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.14683100651636, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 335.1981946982585, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5155438592, "peak_system_memory_used_bytes": 49958428672, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 162.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 148.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 143.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 150.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 148.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 149.47, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7840215230826288}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-06/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-07/seed-1731/checkpoint.json", "checkpoint_sha256": "9880930d3406aa167a0ae34a2685d360722cc3d412feab66de878e334f326824", "completed_utc": "2026-08-02T02:54:01.032888+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.164502748007095, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 336.07253542120156, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5361934336, "peak_system_memory_used_bytes": 52701581312, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 150.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 147.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 141.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 150.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 148.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 143.65, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7547564549604431}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-07/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-07/seed-2731/checkpoint.json", "checkpoint_sha256": "99647501ea1abcc5c469ccc9a521aca7da2519c4fe5f564413a2028d8e2d9b46", "completed_utc": "2026-08-02T02:55:01.985931+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.19670381511819, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 336.1065640627549, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5413560320, "peak_system_memory_used_bytes": 52026994688, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 151.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 149.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 143.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 151.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 148.71, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7697095579933375}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-07/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-07/seed-731/checkpoint.json", "checkpoint_sha256": "079750cd3e4deefbbb69f643695fbcab453d40f26357110c37a42dccfc6062bf", "completed_utc": "2026-08-02T02:53:00.065510+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.182981265039025, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 334.10078660465535, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5310304256, "peak_system_memory_used_bytes": 49959751680, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 153.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 148.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 145.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 145.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 142.82, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7885516640963033}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-07/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-08", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-08/seed-1731/checkpoint.json", "checkpoint_sha256": "3626475c285bf1d131bf7d142313fa7452d5ccff33715c878639b946f8fd2084", "completed_utc": "2026-08-02T02:57:03.885601+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.222176956560645, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 337.2843109168629, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5516816384, "peak_system_memory_used_bytes": 51013541888, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 147.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 144.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 140.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 148.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 147.78, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7654564639087766}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-08/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-08", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-08/seed-2731/checkpoint.json", "checkpoint_sha256": "2d37cc21bc9af9f431563ca8050121b152a5aedb72625d51c9b1a8d8171df172", "completed_utc": "2026-08-02T02:58:04.937262+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.220022917389954, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 334.89148955250647, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5568442368, "peak_system_memory_used_bytes": 43782991872, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 151.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 145.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 144.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 146.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 145.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 147.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7871134700253606}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-08/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-08", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-INV-08/seed-731/checkpoint.json", "checkpoint_sha256": "a6f14641bffebc796ccce9144e2abbfd7e2b9bc82bf65a717ddff5cab27eff27", "completed_utc": "2026-08-02T02:56:02.939079+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.17073377095788, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 338.4823245663107, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5465194496, "peak_system_memory_used_bytes": 55538642944, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 147.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 147.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 152.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 145.31, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 55, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.76318619097583}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/R-INV-08/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-SCHED-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-01/seed-1731/checkpoint.json", "checkpoint_sha256": "e1d6a5cf475394aaeb0c0600c5a1e59b210c464873d0fb88efe58136fd300c59", "completed_utc": "2026-08-02T02:58:28.263999+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 247, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.21408882903603, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 247, "prompt_tokens_per_second": 321.2805866205535, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5660667904, "peak_system_memory_used_bytes": 43748409344, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 159.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 146.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 154.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 138.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 146.24, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.699915750999935}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 189, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "f79dcc7d0c667e80055a13354cb2fa0a186e8972febc22d8e3431278678df2b4", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":2},\"B\":{\"start\":0,\"finish\":1},\"C\":{\"start\":2,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":0,\"finish\":9},\"F\":{\"start\":9,\"finish\":10}},\"makespan\":12}"}]}
{"case_id": "R-SCHED-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-01/seed-2731/checkpoint.json", "checkpoint_sha256": "19458a84898e26d9dc11427b2059317c39b14c30c20c6614e8a0a241f9e7af7e", "completed_utc": "2026-08-02T02:58:43.609034+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 502, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.15796396168622, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 502, "prompt_tokens_per_second": 314.14628437322426, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5699612672, "peak_system_memory_used_bytes": 43943591936, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 156.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 143.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 147.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 145.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 145.38, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6911157229915261}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 189, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "f79dcc7d0c667e80055a13354cb2fa0a186e8972febc22d8e3431278678df2b4", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":2},\"B\":{\"start\":0,\"finish\":1},\"C\":{\"start\":2,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":0,\"finish\":9},\"F\":{\"start\":9,\"finish\":10}},\"makespan\":12}"}]}
{"case_id": "R-SCHED-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-01/seed-731/checkpoint.json", "checkpoint_sha256": "ad172ca0ab49dd1b8e15cb8459a1563929da4e49cc1775e77652e1a244eb44f6", "completed_utc": "2026-08-02T02:58:20.267492+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 500, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.1618577669098, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 500, "prompt_tokens_per_second": 319.7893152746426, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5620027392, "peak_system_memory_used_bytes": 43659309056, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 148.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 140.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 145.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 151.77, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7105283049168065}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 255, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "26ac933ce5dead38a6808254c3024026a1b2d297b954d5c536e289f10c36f6b5", "utf8": "{\n  \"tasks\": {\n    \"A\": {\"start\": 0, \"finish\": 2},\n    \"B\": {\"start\": 0, \"finish\": 1},\n    \"C\": {\"start\": 2, \"finish\": 7},\n    \"D\": {\"start\": 7, \"finish\": 12},\n    \"E\": {\"start\": 0, \"finish\": 9},\n    \"F\": {\"start\": 9, \"finish\": 10}\n  },\n  \"makespan\": 12\n}"}]}
{"case_id": "R-SCHED-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-02/seed-1731/checkpoint.json", "checkpoint_sha256": "57030ef47ddac6b286e196222d9a7aa327c7915c0fd2f256e43a5bd0373afea5", "completed_utc": "2026-08-02T02:59:02.370983+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 284, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.39646092723024, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 284, "prompt_tokens_per_second": 319.1427479919947, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5779578880, "peak_system_memory_used_bytes": 44000083968, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 146.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 137.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 150.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 146.52, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6899237190373242}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-02/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "491ec9d971baf545105f6664d53cff3257eb5b20c6f2c9961563249a470697ab", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":1},\"B\":{\"start\":1,\"finish\":8},\"C\":{\"start\":8,\"finish\":11},\"D\":{\"start\":0,\"finish\":6},\"E\":{\"start\":11,\"finish\":15},\"F\":{\"start\":8,\"finish\":10}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-02/seed-2731/checkpoint.json", "checkpoint_sha256": "728dba39d5129a097c269da64e104fd9e606f1f9a8a4008fb1c6a7c56cbeca75", "completed_utc": "2026-08-02T02:59:10.575131+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 253, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.38587244917512, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 253, "prompt_tokens_per_second": 318.71534166912636, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5818605568, "peak_system_memory_used_bytes": 44049424384, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 155.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 143.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 155.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 145.23, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6898245719494298}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-02/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "491ec9d971baf545105f6664d53cff3257eb5b20c6f2c9961563249a470697ab", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":1},\"B\":{\"start\":1,\"finish\":8},\"C\":{\"start\":8,\"finish\":11},\"D\":{\"start\":0,\"finish\":6},\"E\":{\"start\":11,\"finish\":15},\"F\":{\"start\":8,\"finish\":10}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-02/seed-731/checkpoint.json", "checkpoint_sha256": "0cb917c285508c9de3ef7e411cdaebec01b452f7858f96067897d72042c80e5b", "completed_utc": "2026-08-02T02:58:53.273876+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 303, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.41534248740067, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 303, "prompt_tokens_per_second": 319.154288301659, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5740290048, "peak_system_memory_used_bytes": 43978973184, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 148.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 150.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 147.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 147.5, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.694421058986336}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-02/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "491ec9d971baf545105f6664d53cff3257eb5b20c6f2c9961563249a470697ab", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":1},\"B\":{\"start\":1,\"finish\":8},\"C\":{\"start\":8,\"finish\":11},\"D\":{\"start\":0,\"finish\":6},\"E\":{\"start\":11,\"finish\":15},\"F\":{\"start\":8,\"finish\":10}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-03/seed-1731/checkpoint.json", "checkpoint_sha256": "a4617517cdf1ccd375c98730139f598cb338cc7cb8fde135fbbb5450001fbbfc", "completed_utc": "2026-08-02T02:59:26.955485+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 257, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.354968005817476, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 257, "prompt_tokens_per_second": 318.84433135818216, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5896187904, "peak_system_memory_used_bytes": 44083519488, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 145.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 137.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 145.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 138.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 158.67, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7028232039883733}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 223, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-03/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1bf9185d74873c5fe672817effdbe678e990ac36348da595d4580f673d5d857e", "utf8": "{\"tasks\": {\"A\": {\"start\": 0, \"finish\": 3}, \"B\": {\"start\": 3, \"finish\": 11}, \"C\": {\"start\": 3, \"finish\": 5}, \"D\": {\"start\": 5, \"finish\": 13}, \"E\": {\"start\": 11, \"finish\": 18}, \"F\": {\"start\": 0, \"finish\": 9}}, \"makespan\": 18}"}]}
{"case_id": "R-SCHED-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-03/seed-2731/checkpoint.json", "checkpoint_sha256": "88782e59b9f68e1e7d870978cb113d08e27dbb86dfa539676fcdf12bc04b6af8", "completed_utc": "2026-08-02T02:59:35.040312+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 249, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.374684682278506, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 249, "prompt_tokens_per_second": 319.118852815118, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5935050752, "peak_system_memory_used_bytes": 44233437184, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 145.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 141.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 146.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.27, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6860066240187734}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-03/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fe44bbc412b9a92ffeebd24c82f6f4a272c74e8704837401cc8287b693af26a2", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":3},\"B\":{\"start\":3,\"finish\":11},\"C\":{\"start\":3,\"finish\":5},\"D\":{\"start\":5,\"finish\":13},\"E\":{\"start\":11,\"finish\":18},\"F\":{\"start\":0,\"finish\":9}},\"makespan\":18}"}]}
{"case_id": "R-SCHED-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-03/seed-731/checkpoint.json", "checkpoint_sha256": "69ae35a6132cee4b58ac1452c2b545c0caa9c38dd5257c6cb2cce99073fa6451", "completed_utc": "2026-08-02T02:59:18.625560+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 245, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.33473605024282, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 245, "prompt_tokens_per_second": 319.1944921568225, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5857423360, "peak_system_memory_used_bytes": 44061790208, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 145.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 137.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.93, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6866811600048095}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-03/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fe44bbc412b9a92ffeebd24c82f6f4a272c74e8704837401cc8287b693af26a2", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":3},\"B\":{\"start\":3,\"finish\":11},\"C\":{\"start\":3,\"finish\":5},\"D\":{\"start\":5,\"finish\":13},\"E\":{\"start\":11,\"finish\":18},\"F\":{\"start\":0,\"finish\":9}},\"makespan\":18}"}]}
{"case_id": "R-SCHED-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-04/seed-1731/checkpoint.json", "checkpoint_sha256": "381cf685384206ae460a22f7a02c0307d7641091560d8a7606183387881a2d82", "completed_utc": "2026-08-02T02:59:53.279844+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 309, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.391868735338576, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 309, "prompt_tokens_per_second": 320.9856631674486, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6012792832, "peak_system_memory_used_bytes": 44407033856, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 149.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 146.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 143.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 148.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 147.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.93, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.688927557086572}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-04/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1d34bbff60da03139ba72e62fc0c67dd2d52d5227a9ed2d46b87cf7215f35872", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":9},\"B\":{\"start\":0,\"finish\":6},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":11},\"E\":{\"start\":6,\"finish\":12},\"F\":{\"start\":11,\"finish\":15}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-04/seed-2731/checkpoint.json", "checkpoint_sha256": "5c7d70ba0fe759a647249cd80be98e7a8f2a76b6450bd4e71d8ab116485f74ec", "completed_utc": "2026-08-02T03:00:02.170820+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 268, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.154857804503045, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 268, "prompt_tokens_per_second": 313.1020080172622, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6052016128, "peak_system_memory_used_bytes": 44494073856, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 147.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 147.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 144.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 153.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 150.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 149.91, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7029651999473572}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-04/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1d34bbff60da03139ba72e62fc0c67dd2d52d5227a9ed2d46b87cf7215f35872", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":9},\"B\":{\"start\":0,\"finish\":6},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":11},\"E\":{\"start\":6,\"finish\":12},\"F\":{\"start\":11,\"finish\":15}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-04/seed-731/checkpoint.json", "checkpoint_sha256": "65003f68fc48b337231f3ab59dd161776b349e4f6c9281026dd51c491a950abc", "completed_utc": "2026-08-02T02:59:43.489701+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 262, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.395805488970765, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 262, "prompt_tokens_per_second": 321.24675390441405, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5973856256, "peak_system_memory_used_bytes": 44292935680, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 152.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 140.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.12, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.688526205951348}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-04/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1d34bbff60da03139ba72e62fc0c67dd2d52d5227a9ed2d46b87cf7215f35872", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":9},\"B\":{\"start\":0,\"finish\":6},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":11},\"E\":{\"start\":6,\"finish\":12},\"F\":{\"start\":11,\"finish\":15}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-05/seed-1731/checkpoint.json", "checkpoint_sha256": "b6ba23d6ea12dc076f4ed40aedbbfd71da85bfb65a6dbdd57a0cfa64b489ad22", "completed_utc": "2026-08-02T03:00:30.466876+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 208, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.2309490336896, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 208, "prompt_tokens_per_second": 319.0168623198654, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6132961280, "peak_system_memory_used_bytes": 44736098304, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 145.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 148.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 149.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 149.06, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 149.69, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6932626090710983}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 190, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-05/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "2d4838c5698321ade1f862aa56e5197b4568702edd8457fc104de0df2aa72371", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":3},\"B\":{\"start\":0,\"finish\":8},\"C\":{\"start\":0,\"finish\":4},\"D\":{\"start\":8,\"finish\":13},\"E\":{\"start\":0,\"finish\":8},\"F\":{\"start\":13,\"finish\":20}},\"makespan\":20}"}]}
{"case_id": "R-SCHED-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-05/seed-2731/checkpoint.json", "checkpoint_sha256": "53b01be8989b2376d2bd6bfc038036494ed80c52fa8ae8a7b1d763f274817812", "completed_utc": "2026-08-02T03:00:41.080676+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 338, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.2103175814387, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 338, "prompt_tokens_per_second": 320.9831254585473, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6171484160, "peak_system_memory_used_bytes": 44747251712, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 145.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 140.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 145.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 138.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 143.02, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6845153180183843}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 222, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-05/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "17661ecb9fa54b3edf018690fd2284b13649ea1cb4ad1dfd400f5599dbfc7b16", "utf8": "{\"tasks\": {\"A\": {\"start\": 0, \"finish\": 3}, \"B\": {\"start\": 0, \"finish\": 8}, \"C\": {\"start\": 0, \"finish\": 4}, \"D\": {\"start\": 8, \"finish\": 13}, \"E\": {\"start\": 0, \"finish\": 8}, \"F\": {\"start\": 13, \"finish\": 20}}, \"makespan\": 20}"}]}
{"case_id": "R-SCHED-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-05/seed-731/checkpoint.json", "checkpoint_sha256": "804fe30e91a94cedc1b6ae9a2f74eaadb14360ac66d982fff4e00f8ad0661b93", "completed_utc": "2026-08-02T03:00:23.462755+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 708, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.97423934857811, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 708, "prompt_tokens_per_second": 321.44463234968106, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6090956800, "peak_system_memory_used_bytes": 45719355392, "per_gpu": {"0": {"peak_memory_used_mib": 17892.0, "peak_power_draw_w": 150.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16840.0, "peak_power_draw_w": 145.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16810.0, "peak_power_draw_w": 146.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19380.0, "peak_power_draw_w": 150.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 145.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 146.3, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 19, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6749428389593959}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 190, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-05/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "2d4838c5698321ade1f862aa56e5197b4568702edd8457fc104de0df2aa72371", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":3},\"B\":{\"start\":0,\"finish\":8},\"C\":{\"start\":0,\"finish\":4},\"D\":{\"start\":8,\"finish\":13},\"E\":{\"start\":0,\"finish\":8},\"F\":{\"start\":13,\"finish\":20}},\"makespan\":20}"}]}
{"case_id": "R-SCHED-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-06/seed-1731/checkpoint.json", "checkpoint_sha256": "30d9ed27e43b24310d6fea70291c8d278a4f6054b2be6eef1c14bd6296c7b7ac", "completed_utc": "2026-08-02T03:00:56.358377+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 218, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.207690651663675, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 218, "prompt_tokens_per_second": 320.29429019931183, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6249730048, "peak_system_memory_used_bytes": 44744384512, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 144.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 142.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 141.86, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6761749190045521}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-06/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8f2df7d48aa132b0a78bc8a0a5d48922df0d7c7ef15551d86c790ff2acbff5fc", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":7},\"B\":{\"start\":0,\"finish\":2},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":2,\"finish\":10},\"F\":{\"start\":10,\"finish\":18}},\"makespan\":18}"}]}
{"case_id": "R-SCHED-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-06/seed-2731/checkpoint.json", "checkpoint_sha256": "e282e5a2ec51df27166b2e78429d9fb0e4250f87e5fd19a50585dd1464c6272e", "completed_utc": "2026-08-02T03:01:03.940744+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 231, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.26685790158531, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 231, "prompt_tokens_per_second": 317.9279908839657, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6288297984, "peak_system_memory_used_bytes": 44787335168, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 145.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 139.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 151.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 152.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 140.84, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6904155641095713}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-06/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8f2df7d48aa132b0a78bc8a0a5d48922df0d7c7ef15551d86c790ff2acbff5fc", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":7},\"B\":{\"start\":0,\"finish\":2},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":2,\"finish\":10},\"F\":{\"start\":10,\"finish\":18}},\"makespan\":18}"}]}
{"case_id": "R-SCHED-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-06/seed-731/checkpoint.json", "checkpoint_sha256": "9fae88ec46926b8a8d488e7c0f58d08f1b7156826eb65c6cd61e3695d56a6b91", "completed_utc": "2026-08-02T03:00:49.092090+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 245, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.24209304085177, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 245, "prompt_tokens_per_second": 317.9650238473768, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6210936832, "peak_system_memory_used_bytes": 44724006912, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 152.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 147.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 138.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 153.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 152.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 150.55, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6887132230913267}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-06/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8f2df7d48aa132b0a78bc8a0a5d48922df0d7c7ef15551d86c790ff2acbff5fc", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":7},\"B\":{\"start\":0,\"finish\":2},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":2,\"finish\":10},\"F\":{\"start\":10,\"finish\":18}},\"makespan\":18}"}]}
{"case_id": "R-SCHED-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-07/seed-1731/checkpoint.json", "checkpoint_sha256": "9b33e23c75f3a820d2e0283c7cdc2a4f3241364ebc1d8a625406a2b64d9384d8", "completed_utc": "2026-08-02T03:01:32.790009+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 514, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.895627788298135, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 514, "prompt_tokens_per_second": 320.4297528449921, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6367023104, "peak_system_memory_used_bytes": 44987408384, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 146.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 142.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 140.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 144.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 139.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 142.17, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.696109844953753}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-07/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "6abc19a4f7733697224a9ff2a40ce7c618de36f1b51b1785ea8c07344fa7d46f", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":9},\"B\":{\"start\":0,\"finish\":9},\"C\":{\"start\":0,\"finish\":4},\"D\":{\"start\":9,\"finish\":10},\"E\":{\"start\":9,\"finish\":10},\"F\":{\"start\":10,\"finish\":13}},\"makespan\":13}"}]}
{"case_id": "R-SCHED-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-07/seed-2731/checkpoint.json", "checkpoint_sha256": "d0c39183876f8620dd84fb4a9b919b877c9d05e6ec769125e5750d15c59b5ae7", "completed_utc": "2026-08-02T03:01:47.011381+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 460, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.17670213370405, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 460, "prompt_tokens_per_second": 314.09259572895655, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6407684096, "peak_system_memory_used_bytes": 44996538368, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 154.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 139.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 155.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 144.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 144.15, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7084936250466853}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 223, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-07/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "4425e2339ccff33b11d4fd0c7da4fd638c12402debea13ee33c2270971058599", "utf8": "{\"tasks\": {\"A\": {\"start\": 0, \"finish\": 9}, \"B\": {\"start\": 0, \"finish\": 9}, \"C\": {\"start\": 0, \"finish\": 4}, \"D\": {\"start\": 9, \"finish\": 10}, \"E\": {\"start\": 9, \"finish\": 10}, \"F\": {\"start\": 10, \"finish\": 13}}, \"makespan\": 13}"}]}
{"case_id": "R-SCHED-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/R-SCHED-07/seed-731/checkpoint.json", "checkpoint_sha256": "c2339822e54f197623bf16f731eb2641e51d37546b12ac7490388d271954010c", "completed_utc": "2026-08-02T03:01:16.962899+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 420, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.20397476344396, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 420, "prompt_tokens_per_second": 321.1597013844502, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6327021568, "peak_system_memory_used_bytes": 44960022528, "per_gpu": {"0": {"peak_memory_used_mib": 17625.0, "peak_power_draw_w": 151.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 141.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16543.0, "peak_power_draw_w": 141.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19113.0, "peak_power_draw_w": 143.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16573.0, "peak_power_draw_w": 143.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15153.0, "peak_power_draw_w": 152.29, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 11, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6806723580230027}, "quant": "UD-IQ2_M", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ2_M/cases/R-SCHED-07/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "6abc19a4f7733697224a9ff2a40ce7c618de36f1b51b1785ea8c07344fa7d46f", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":9},\"B\":{\"start\":0,\"finish\":9},\"C\":{\"start\":0,\"finish\":4},\"D\":{\"start\":9,\"finish\":10},\"E\":{\"start\":9,\"finish\":10},\"F\":{\"start\":10,\"finish\":13}},\"makespan\":13}"}]}
{"case_id": "STRICT-JSON-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/STRICT-JSON-01/seed-1731/checkpoint.json", "checkpoint_sha256": "368c2069d3b3be51de8c18a1324b1c0d2172e982a936afbbc9af5316852a7455", "completed_utc": "2026-08-02T03:04:05.748680+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": [989, 361, 582, 803, 4628, 17462, 6288, 3362, 5062, 201, 223, 7413, 223, 875, 361, 582, 86, 29929, 5165, 3362], "id": "repeated_20_token_ngram", "passed": false}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 1739, "passed": true, "task_completion_valid": true}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.34378578249352, "finish_reason": "stop", "output_token_cap": 4096, "output_tokens": 1739, "prompt_tokens_per_second": 750.0416531850613, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6512300032, "peak_system_memory_used_bytes": 54591434752, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 171.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 166.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 148.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 150.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 172.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 155.99, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 48, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.8296477670082822}, "quant": "UD-IQ2_M", "score": {"components": {"exact_shape": 1.0, "factual_slots": 6.0, "parse": 1.0, "schema_validity": 2.0}, "factual_slots_correct": 72, "factual_slots_total": 72, "max_score": 10.0, "parse_error": null, "passed": true, "score": 10.0, "scorer": "strict_json_ajv_equivalent"}, "seed": 1731, "status": "completed", "task": "strict_json", "visible_model_outputs": [{"bytes": 2184, "path": "results/quants/UD-IQ2_M/cases/STRICT-JSON-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9a4b44f10dbeb47ff6300d9ad5e7d61929b997496b1ec630e147227c0a28e94e", "utf8": "[\n  {\n    \"ticket_id\": \"T-4107\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-07-31T22:14:00-07:00\",\n    \"customer\": \"Alma\",\n    \"refund_cents\": 1840,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4108\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T09:02:11+02:00\",\n    \"customer\": \"Benoît\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4109\",\n    \"priority\": \"low\",\n    \"opened_at\": \"2026-08-01T17:33:20+09:00\",\n    \"customer\": \"Chika\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4110\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T08:05:00-03:00\",\n    \"customer\": \"Diego\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4111\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T12:00:00Z\",\n    \"customer\": \"Eleni\",\n    \"refund_cents\": 10299,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4112\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T15:45:30+04:00\",\n    \"customer\": \"Faisal\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4113\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-08-01T07:18:44-05:00\",\n    \"customer\": \"Grace\",\n    \"refund_cents\": 2400,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4114\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-02T00:03:09+09:00\",\n    \"customer\": \"Haruto\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4115\",\n    \"priority\": \"low\",\n    \"opened_at\": \"2026-08-01T13:27:00+01:00\",\n    \"customer\": \"Inês\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4116\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-08-01T10:11:12-04:00\",\n    \"customer\": \"Jamal\",\n    \"refund_cents\": 5550,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4117\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T16:40:05+02:00\",\n    \"customer\": \"Klara\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4118\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T18:22:31Z\",\n    \"customer\": \"Luis\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  }\n]"}]}
{"case_id": "STRICT-JSON-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/STRICT-JSON-01/seed-2731/checkpoint.json", "checkpoint_sha256": "cba350953ae8ca026cc12239315b15ff748fc50c3a7a93402d78943d4b567a30", "completed_utc": "2026-08-02T03:05:03.737068+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": [989, 361, 582, 803, 4628, 17462, 6288, 3362, 5062, 201, 223, 7413, 223, 875, 361, 582, 86, 29929, 5165, 3362], "id": "repeated_20_token_ngram", "passed": false}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 1860, "passed": true, "task_completion_valid": true}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.36900108782943, "finish_reason": "stop", "output_token_cap": 4096, "output_tokens": 1860, "prompt_tokens_per_second": 746.6319483754104, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6569283584, "peak_system_memory_used_bytes": 50710306816, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 174.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 149.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 142.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 170.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 164.41, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 52, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.853359289932996}, "quant": "UD-IQ2_M", "score": {"components": {"exact_shape": 1.0, "factual_slots": 6.0, "parse": 1.0, "schema_validity": 2.0}, "factual_slots_correct": 72, "factual_slots_total": 72, "max_score": 10.0, "parse_error": null, "passed": true, "score": 10.0, "scorer": "strict_json_ajv_equivalent"}, "seed": 2731, "status": "completed", "task": "strict_json", "visible_model_outputs": [{"bytes": 2184, "path": "results/quants/UD-IQ2_M/cases/STRICT-JSON-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9a4b44f10dbeb47ff6300d9ad5e7d61929b997496b1ec630e147227c0a28e94e", "utf8": "[\n  {\n    \"ticket_id\": \"T-4107\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-07-31T22:14:00-07:00\",\n    \"customer\": \"Alma\",\n    \"refund_cents\": 1840,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4108\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T09:02:11+02:00\",\n    \"customer\": \"Benoît\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4109\",\n    \"priority\": \"low\",\n    \"opened_at\": \"2026-08-01T17:33:20+09:00\",\n    \"customer\": \"Chika\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4110\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T08:05:00-03:00\",\n    \"customer\": \"Diego\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4111\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T12:00:00Z\",\n    \"customer\": \"Eleni\",\n    \"refund_cents\": 10299,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4112\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T15:45:30+04:00\",\n    \"customer\": \"Faisal\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4113\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-08-01T07:18:44-05:00\",\n    \"customer\": \"Grace\",\n    \"refund_cents\": 2400,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4114\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-02T00:03:09+09:00\",\n    \"customer\": \"Haruto\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4115\",\n    \"priority\": \"low\",\n    \"opened_at\": \"2026-08-01T13:27:00+01:00\",\n    \"customer\": \"Inês\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4116\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-08-01T10:11:12-04:00\",\n    \"customer\": \"Jamal\",\n    \"refund_cents\": 5550,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4117\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T16:40:05+02:00\",\n    \"customer\": \"Klara\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4118\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T18:22:31Z\",\n    \"customer\": \"Luis\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  }\n]"}]}
{"case_id": "STRICT-JSON-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/STRICT-JSON-01/seed-731/checkpoint.json", "checkpoint_sha256": "62896f407a20097d9796ace058d2b3da76a12f9960778101d25aeab03e0acdff", "completed_utc": "2026-08-02T03:03:11.384697+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": [989, 361, 582, 803, 4628, 17462, 6288, 3362, 5062, 201, 223, 7413, 223, 875, 361, 582, 86, 29929, 5165, 3362], "id": "repeated_20_token_ngram", "passed": false}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": false, "task_completion_valid": false}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 2735, "passed": true, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.238722681704516, "finish_reason": "stop", "output_token_cap": 4096, "output_tokens": 2735, "prompt_tokens_per_second": 746.5778546556119, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6448279552, "peak_system_memory_used_bytes": 48856715264, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 171.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 145.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 159.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 168.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 156.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 154.91, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 76, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.8334924159571528}, "quant": "UD-IQ2_M", "score": {"components": {"exact_shape": 0.0, "factual_slots": 0.0, "parse": 0.0, "schema_validity": 0.0}, "max_score": 10.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "score": 0.0, "scorer": "strict_json_ajv_equivalent"}, "seed": 731, "status": "completed", "task": "strict_json", "visible_model_outputs": [{"bytes": 2196, "path": "results/quants/UD-IQ2_M/cases/STRICT-JSON-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "f682ab1bc1e424e44fca75a229f2868c94c7cca91d62807634af576875ecd6ca", "utf8": "```json\n[\n  {\n    \"ticket_id\": \"T-4107\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-07-31T22:14:00-07:00\",\n    \"customer\": \"Alma\",\n    \"refund_cents\": 1840,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4108\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T09:02:11+02:00\",\n    \"customer\": \"Benoît\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4109\",\n    \"priority\": \"low\",\n    \"opened_at\": \"2026-08-01T17:33:20+09:00\",\n    \"customer\": \"Chika\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4110\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T08:05:00-03:00\",\n    \"customer\": \"Diego\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4111\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T12:00:00Z\",\n    \"customer\": \"Eleni\",\n    \"refund_cents\": 10299,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4112\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T15:45:30+04:00\",\n    \"customer\": \"Faisal\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4113\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-08-01T07:18:44-05:00\",\n    \"customer\": \"Grace\",\n    \"refund_cents\": 2400,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4114\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-02T00:03:09+09:00\",\n    \"customer\": \"Haruto\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4115\",\n    \"priority\": \"low\",\n    \"opened_at\": \"2026-08-01T13:27:00+01:00\",\n    \"customer\": \"Inês\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4116\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-08-01T10:11:12-04:00\",\n    \"customer\": \"Jamal\",\n    \"refund_cents\": 5550,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4117\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T16:40:05+02:00\",\n    \"customer\": \"Klara\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4118\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T18:22:31Z\",\n    \"customer\": \"Luis\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  }\n]\n```"}]}
{"case_id": "TC-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-01/seed-1731/checkpoint.json", "checkpoint_sha256": "50b04489e0b04af5a8f405528ba23015dad7df5e1ece9630083d436f50776825", "completed_utc": "2026-08-02T03:05:16.973032+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 107, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.397225590507134, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 107, "prompt_tokens_per_second": 639.5346155797397, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6627016704, "peak_system_memory_used_bytes": 50432864256, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 164.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 152.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 152.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 150.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 159.81, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3346292909700423}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"amount_cents": 1840, "invoice_id": "INV-8841", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 121, "path": "results/quants/UD-IQ2_M/cases/TC-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ea3c45ab80c1244f2974490a94b75690071f54a20bb2ef2af0d0659d6a2e9e65", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-8841\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-01/seed-2731/checkpoint.json", "checkpoint_sha256": "306144d59ed0dc524f2030b65962549159f573a675f1dec6f27a92e365670c09", "completed_utc": "2026-08-02T03:05:20.836519+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 76, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.64742019473887, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 76, "prompt_tokens_per_second": 643.1598940332232, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6627192832, "peak_system_memory_used_bytes": 50455830528, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 170.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 146.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 149.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 150.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 153.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 139.81, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3175690340576693}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"amount_cents": 1840, "invoice_id": "INV-8841", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 121, "path": "results/quants/UD-IQ2_M/cases/TC-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ea3c45ab80c1244f2974490a94b75690071f54a20bb2ef2af0d0659d6a2e9e65", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-8841\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-01/seed-731/checkpoint.json", "checkpoint_sha256": "d944070245d60a85da357313b630514035d1dd42dd7a726c1c70b423ac86fafd", "completed_utc": "2026-08-02T03:05:12.264993+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 229, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.10734669268953, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 229, "prompt_tokens_per_second": 641.1214694010907, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6626992128, "peak_system_memory_used_bytes": 50332131328, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 166.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 147.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 152.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 154.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 155.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 163.76, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3774727049749345}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"amount_cents": 1840, "invoice_id": "INV-8841", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 121, "path": "results/quants/UD-IQ2_M/cases/TC-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ea3c45ab80c1244f2974490a94b75690071f54a20bb2ef2af0d0659d6a2e9e65", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-8841\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-02/seed-1731/checkpoint.json", "checkpoint_sha256": "9b98db6991b128c9a1d765f67ae5a4d87118a80ee3963c58fb44112fd3af43c3", "completed_utc": "2026-08-02T03:05:39.871860+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 96, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.84340515420959, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 96, "prompt_tokens_per_second": 679.5524712091791, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6627889152, "peak_system_memory_used_bytes": 50599370752, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 163.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 159.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 156.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 161.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 166.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 145.44, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4550860850140452}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "lee@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"amount_cents": 1840, "invoice_id": "INV-9920", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ2_M/cases/TC-02/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcc872a46c28b70c804a2802d00324e1ecc112dd9defb810ff2369d74e4038a2", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"lee@example.test\"}}"}, {"bytes": 121, "path": "results/quants/UD-IQ2_M/cases/TC-02/seed-1731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "e5f900c628fee389011b5077910c06588f7a30f594d669fad58825ed813c431d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-9920\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-02/seed-2731/checkpoint.json", "checkpoint_sha256": "869a9ed03dcf7b1b58357628490298cd485912b7f4501437923c3c696f75a999", "completed_utc": "2026-08-02T03:05:51.165336+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 137, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.52312643039143, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 137, "prompt_tokens_per_second": 680.4975492174632, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6628229120, "peak_system_memory_used_bytes": 48950128640, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 177.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 166.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 154.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 156.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 151.04, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.44561584899202}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "lee@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"amount_cents": 1840, "invoice_id": "INV-9920", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ2_M/cases/TC-02/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcc872a46c28b70c804a2802d00324e1ecc112dd9defb810ff2369d74e4038a2", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"lee@example.test\"}}"}, {"bytes": 121, "path": "results/quants/UD-IQ2_M/cases/TC-02/seed-2731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "e5f900c628fee389011b5077910c06588f7a30f594d669fad58825ed813c431d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-9920\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-02", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-02/seed-731/checkpoint.json", "checkpoint_sha256": "10ee541bfb72da101a6007e223b584c06d24a760191a0441545a2eb694ce0502", "completed_utc": "2026-08-02T03:05:29.100335+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 99, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.93804763667777, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 99, "prompt_tokens_per_second": 681.6429510245569, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6627577856, "peak_system_memory_used_bytes": 50522198016, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 187.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 147.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 139.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 161.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 157.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 156.1, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4417756519978866}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "lee@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"amount_cents": 1840, "invoice_id": "INV-9920", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ2_M/cases/TC-02/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcc872a46c28b70c804a2802d00324e1ecc112dd9defb810ff2369d74e4038a2", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"lee@example.test\"}}"}, {"bytes": 121, "path": "results/quants/UD-IQ2_M/cases/TC-02/seed-731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "e5f900c628fee389011b5077910c06588f7a30f594d669fad58825ed813c431d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-9920\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-03/seed-1731/checkpoint.json", "checkpoint_sha256": "b18a9096c597116d66b03fbea7c777dd31de0c687e94322b0746b3e8f4aa328a", "completed_utc": "2026-08-02T03:05:59.615333+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 87, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.25295293919864, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 87, "prompt_tokens_per_second": 646.5353900789299, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6628589568, "peak_system_memory_used_bytes": 48924344320, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 185.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 143.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 153.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 152.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 155.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 154.18, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.316703912918456}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-117", "note": "Deceased-owner access request.", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 135, "path": "results/quants/UD-IQ2_M/cases/TC-03/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b3016a3eb67c2409660d39dc9bf8f3c733ba57dd6feba22859434bde5d41bf03", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-117\",\"queue\":\"legal\",\"note\":\"Deceased-owner access request.\"}}"}]}
{"case_id": "TC-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-03/seed-2731/checkpoint.json", "checkpoint_sha256": "8d263fcc7288c6e8933da61350e84c74aa7515c9b69b073d913bb5fa4cf51d9d", "completed_utc": "2026-08-02T03:06:04.291483+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 103, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.342892104969216, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 103, "prompt_tokens_per_second": 645.6849937709934, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6628769792, "peak_system_memory_used_bytes": 49129177088, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 157.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 155.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 143.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 168.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 155.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 139.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3257883579935879}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-117", "note": "Deceased-owner access request.", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 135, "path": "results/quants/UD-IQ2_M/cases/TC-03/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b3016a3eb67c2409660d39dc9bf8f3c733ba57dd6feba22859434bde5d41bf03", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-117\",\"queue\":\"legal\",\"note\":\"Deceased-owner access request.\"}}"}]}
{"case_id": "TC-03", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-03/seed-731/checkpoint.json", "checkpoint_sha256": "c6a7b3f50b22b18ef6a2a663f10555a9274a000f74582ffe63b57e3174036226", "completed_utc": "2026-08-02T03:05:55.357533+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 89, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.31781769526935, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 89, "prompt_tokens_per_second": 646.7539681373099, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6628413440, "peak_system_memory_used_bytes": 49006563328, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 167.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 142.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 151.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 149.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 152.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 172.67, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3167630030075088}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-117", "note": "Deceased-owner access request.", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 135, "path": "results/quants/UD-IQ2_M/cases/TC-03/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b3016a3eb67c2409660d39dc9bf8f3c733ba57dd6feba22859434bde5d41bf03", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-117\",\"queue\":\"legal\",\"note\":\"Deceased-owner access request.\"}}"}]}
{"case_id": "TC-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-04/seed-1731/checkpoint.json", "checkpoint_sha256": "7f42326a1e4cd3c38890a029cc58fbf784082c51529ec046ca7eb6c4e896a830", "completed_utc": "2026-08-02T03:06:15.027674+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 137, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.30956139858795, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 137, "prompt_tokens_per_second": 639.0091581035072, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6629130240, "peak_system_memory_used_bytes": 52359290880, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 167.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 149.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 151.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 160.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 164.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 153.57, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3230758929857984}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"kind": "ask_user", "message": "Please provide the invoice identifier (INV-...) for the $9.00 refund."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 101, "path": "results/quants/UD-IQ2_M/cases/TC-04/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "501b59e5f2fae8ee217cfa8541be26873ad79f5e1a5456bc1158718adb946912", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Please provide the invoice identifier (INV-...) for the $9.00 refund.\"}"}]}
{"case_id": "TC-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-04/seed-2731/checkpoint.json", "checkpoint_sha256": "572d17d48ebcd190c7881ce841f733fff8bc267bbdbb4c3d9fac49b5f21d7590", "completed_utc": "2026-08-02T03:06:21.087839+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 147, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.818502309665675, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 147, "prompt_tokens_per_second": 636.3144421886453, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6629302272, "peak_system_memory_used_bytes": 53939650560, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 163.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 156.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 151.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 164.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 158.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 145.72, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3276641310658306}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"kind": "ask_user", "message": "Please provide the invoice identifier, customer identifier, or customer email so I can locate the invoice for the $9.00 refund."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 159, "path": "results/quants/UD-IQ2_M/cases/TC-04/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a15f3381fafd5eed1ff3899fe1224648cf798d2fa579598a7acbe6440b782793", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Please provide the invoice identifier, customer identifier, or customer email so I can locate the invoice for the $9.00 refund.\"}"}]}
{"case_id": "TC-04", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-04/seed-731/checkpoint.json", "checkpoint_sha256": "c821f074f937778152b3fed9eb18f770a565554cc42c18f1c0c189547a46f836", "completed_utc": "2026-08-02T03:06:09.329644+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 115, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.27181482968249, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 115, "prompt_tokens_per_second": 639.042653977289, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6628950016, "peak_system_memory_used_bytes": 49016307712, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 150.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 155.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 141.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 150.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 150.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 140.89, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.322882543085143}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"kind": "ask_user", "message": "What is the customer, email, or invoice identifier for the $9.00 refund?"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 104, "path": "results/quants/UD-IQ2_M/cases/TC-04/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ec1e708581bd21743d79f69134577d0e9d0bab6c05fd600ab2a0138e4a8b9c3a", "utf8": "{\"kind\":\"ask_user\",\"message\":\"What is the customer, email, or invoice identifier for the $9.00 refund?\"}"}]}
{"case_id": "TC-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-05/seed-1731/checkpoint.json", "checkpoint_sha256": "33a9a97dc22d804f3b5292770f28894dcb9e693e64c495b38696cdeed8079180", "completed_utc": "2026-08-02T03:06:27.460291+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 53, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.900750168577204, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 53, "prompt_tokens_per_second": 636.0193831538226, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6629650432, "peak_system_memory_used_bytes": 53136666624, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 169.36, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 153.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 149.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 159.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 150.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 139.72, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3282137360656634}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-3310"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-05/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "620e7956cad16843a07c4a9447ddc6b4480d9c8e414c5524ebcfcef65dad591d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-3310\"}}"}]}
{"case_id": "TC-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-05/seed-2731/checkpoint.json", "checkpoint_sha256": "0dd09c87c1ec6b9bcbc8da9ae027ca23b4b2a261342c69726facd2df25cd3a82", "completed_utc": "2026-08-02T03:06:30.371883+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 46, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.01035469294778, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 46, "prompt_tokens_per_second": 635.0444338432442, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6629814272, "peak_system_memory_used_bytes": 50032459776, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 161.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 150.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 154.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 153.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 146.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 156.83, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3223897350253537}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-3310"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-05/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "620e7956cad16843a07c4a9447ddc6b4480d9c8e414c5524ebcfcef65dad591d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-3310\"}}"}]}
{"case_id": "TC-05", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-05/seed-731/checkpoint.json", "checkpoint_sha256": "eb11d89771cdf61bda2748efbae7ac98e1c024a5b88ba718bd26de11d1e53061", "completed_utc": "2026-08-02T03:06:24.218443+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 50, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.90423325521766, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 50, "prompt_tokens_per_second": 636.0419664194542, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6629486592, "peak_system_memory_used_bytes": 53169352704, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 156.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 153.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 141.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 160.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 150.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 152.45, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3274952929932624}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-3310"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-05/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "620e7956cad16843a07c4a9447ddc6b4480d9c8e414c5524ebcfcef65dad591d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-3310\"}}"}]}
{"case_id": "TC-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-06/seed-1731/checkpoint.json", "checkpoint_sha256": "9732678dc4d267c589bc7f2e896b71e3d64213da6549704e4eff26b4d0d85f49", "completed_utc": "2026-08-02T03:06:45.949818+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 133, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.104732813778924, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 133, "prompt_tokens_per_second": 669.6731680472273, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6630457344, "peak_system_memory_used_bytes": 53664681984, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 151.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 153.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 150.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 164.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 168.58, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4329655229812488}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "amy@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-606"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ2_M/cases/TC-06/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "5f1f41cbae0966614e1c06469d63fedbfcad8f31107fe71236878b61cc785043", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"amy@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ2_M/cases/TC-06/seed-1731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "2519b5e57415d9a165fbf2aac2692445aa6f524f1b49f31c3f864a483728e054", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-606\"}}"}]}
{"case_id": "TC-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-06/seed-2731/checkpoint.json", "checkpoint_sha256": "2a0b2863db476a193a134b42c09cfd1de206924a2a68fe47b1a3e946199dcd6a", "completed_utc": "2026-08-02T03:06:56.362451+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 152, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.80670128888049, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 152, "prompt_tokens_per_second": 670.9784971800284, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6630797312, "peak_system_memory_used_bytes": 54941491200, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 155.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 148.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 150.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 162.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 151.74, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4275872519938275}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "amy@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-606"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ2_M/cases/TC-06/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "5f1f41cbae0966614e1c06469d63fedbfcad8f31107fe71236878b61cc785043", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"amy@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ2_M/cases/TC-06/seed-2731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "2519b5e57415d9a165fbf2aac2692445aa6f524f1b49f31c3f864a483728e054", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-606\"}}"}]}
{"case_id": "TC-06", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-06/seed-731/checkpoint.json", "checkpoint_sha256": "e54f283455f55946705e77ae00881f8801eacbe66a584dace25afd90601c8d59", "completed_utc": "2026-08-02T03:06:37.440789+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 96, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.98070108557721, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 96, "prompt_tokens_per_second": 671.9119300580554, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6630141952, "peak_system_memory_used_bytes": 54143602688, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 172.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 158.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 153.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 178.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 160.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 158.11, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4261951580410823}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "amy@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-606"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ2_M/cases/TC-06/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "5f1f41cbae0966614e1c06469d63fedbfcad8f31107fe71236878b61cc785043", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"amy@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ2_M/cases/TC-06/seed-731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "2519b5e57415d9a165fbf2aac2692445aa6f524f1b49f31c3f864a483728e054", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-606\"}}"}]}
{"case_id": "TC-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-07/seed-1731/checkpoint.json", "checkpoint_sha256": "1398854cecac7b0dba5c1b3e46c2d26a35ab68459e8c2de0c0a834875eb3e473", "completed_utc": "2026-08-02T03:07:02.858823+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 56, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.94589937952308, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 56, "prompt_tokens_per_second": 634.9814526707072, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6631124992, "peak_system_memory_used_bytes": 53402370048, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 158.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 141.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 151.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 157.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 158.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 151.97, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3204064139863476}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-808"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 78, "path": "results/quants/UD-IQ2_M/cases/TC-07/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ae97c01bf7ef4a65020b0039d0f8cc2ecaf3668fff9be418d02cd00c64383426", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-808\"}}"}]}
{"case_id": "TC-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-07/seed-2731/checkpoint.json", "checkpoint_sha256": "dd9a4a155fbfbe96f8fc5f92bf7ad28bb6d1ab9c8703d8b64c1c7f180dab8059", "completed_utc": "2026-08-02T03:07:06.638548+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 70, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.42410198582809, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 70, "prompt_tokens_per_second": 635.271969049426, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6631297024, "peak_system_memory_used_bytes": 53328461824, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 162.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 152.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 138.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 162.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 152.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 138.72, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3200960659887642}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-808"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 78, "path": "results/quants/UD-IQ2_M/cases/TC-07/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ae97c01bf7ef4a65020b0039d0f8cc2ecaf3668fff9be418d02cd00c64383426", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-808\"}}"}]}
{"case_id": "TC-07", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-07/seed-731/checkpoint.json", "checkpoint_sha256": "d330628ecc4c1f2feb419622bdbd8536d727e66262a7e1be90e3e93f23336039", "completed_utc": "2026-08-02T03:06:59.435602+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 47, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.9231282503587, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 47, "prompt_tokens_per_second": 635.538181773198, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6630969344, "peak_system_memory_used_bytes": 53785788416, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 148.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 150.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 148.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 155.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 142.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 147.88, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3183539720484987}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-808"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 78, "path": "results/quants/UD-IQ2_M/cases/TC-07/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ae97c01bf7ef4a65020b0039d0f8cc2ecaf3668fff9be418d02cd00c64383426", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-808\"}}"}]}
{"case_id": "TC-08", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-08/seed-1731/checkpoint.json", "checkpoint_sha256": "212352827ad63f39f83efe25887e58202477d157d5c43545e94b79dc5871f0be", "completed_utc": "2026-08-02T03:07:18.290616+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 119, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.308413028317126, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 119, "prompt_tokens_per_second": 641.046485160703, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6631657472, "peak_system_memory_used_bytes": 53884829696, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 152.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 142.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 151.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 149.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 152.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 154.28, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3012725219596177}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"amount_cents": 2200, "invoice_id": "INV-7001", "reason": "fraud"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 117, "path": "results/quants/UD-IQ2_M/cases/TC-08/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e6d61ba2e18b0b4002df102f492835cd369c3caa7ce725f57fa20c1809daf77e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-7001\",\"amount_cents\":2200,\"reason\":\"fraud\"}}"}]}
{"case_id": "TC-08", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-08/seed-2731/checkpoint.json", "checkpoint_sha256": "74f48ae7c0c915e7de1d8826d3bb05982f3077073c8c9dea3e46dad0f8c9ab2a", "completed_utc": "2026-08-02T03:07:23.175060+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 111, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.38314470099984, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 111, "prompt_tokens_per_second": 637.8243718238773, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6631829504, "peak_system_memory_used_bytes": 53540884480, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 166.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 142.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 142.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 165.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 151.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 142.64, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3233409239910543}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"amount_cents": 2200, "invoice_id": "INV-7001", "reason": "fraud"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 117, "path": "results/quants/UD-IQ2_M/cases/TC-08/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e6d61ba2e18b0b4002df102f492835cd369c3caa7ce725f57fa20c1809daf77e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-7001\",\"amount_cents\":2200,\"reason\":\"fraud\"}}"}]}
{"case_id": "TC-08", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-08/seed-731/checkpoint.json", "checkpoint_sha256": "f13d056723f8ff67b1c4a4d21dddddd860cd38b74cab597d7098da550a62ec46", "completed_utc": "2026-08-02T03:07:13.170075+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 169, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.421213406838575, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 169, "prompt_tokens_per_second": 638.6296919614416, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6631477248, "peak_system_memory_used_bytes": 54016417792, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 151.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 152.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 149.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 150.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 152.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 165.84, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3118559470167384}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"amount_cents": 2200, "invoice_id": "INV-7001", "reason": "fraud"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 117, "path": "results/quants/UD-IQ2_M/cases/TC-08/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e6d61ba2e18b0b4002df102f492835cd369c3caa7ce725f57fa20c1809daf77e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-7001\",\"amount_cents\":2200,\"reason\":\"fraud\"}}"}]}
{"case_id": "TC-09", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-09/seed-1731/checkpoint.json", "checkpoint_sha256": "b794286a7d857af2f009a84a8763b882e70564ec8ea0f144c60f2eed930036eb", "completed_utc": "2026-08-02T03:07:32.664363+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 137, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.19813852292998, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 137, "prompt_tokens_per_second": 642.7273347516234, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6632161280, "peak_system_memory_used_bytes": 52888281088, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 166.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 156.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 137.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 168.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 158.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 146.28, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3148992740316316}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-7440"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-09/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d6c84e227bb66da4c0cdd7cab54902164e489bc38b561066cbda2d965f3bc42e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-7440\"}}"}]}
{"case_id": "TC-09", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-09/seed-2731/checkpoint.json", "checkpoint_sha256": "ffafb3f1c05a4df84c96e3acf0d8e91598df5e4afdcd2b6ca46d7d087722b55e", "completed_utc": "2026-08-02T03:07:38.207853+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 135, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.89192242515534, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 135, "prompt_tokens_per_second": 639.8925536267011, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6632341504, "peak_system_memory_used_bytes": 53647953920, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 161.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 145.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 151.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 151.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 160.18, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3087081239791587}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"amount_cents": 6500, "invoice_id": "INV-7440", "reason": "service_failure"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 127, "path": "results/quants/UD-IQ2_M/cases/TC-09/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "322e392693a24fe4431bb53ad571a7dc8da4c0d4a3e509f5b3678bf957db1739", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-7440\",\"amount_cents\":6500,\"reason\":\"service_failure\"}}"}]}
{"case_id": "TC-09", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-09/seed-731/checkpoint.json", "checkpoint_sha256": "35c0aac8b4528578a32e796a5ecaa767ef1cb308f400b825ac96b39fd1fa51c5", "completed_utc": "2026-08-02T03:07:27.028440+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 77, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.10737888779824, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 77, "prompt_tokens_per_second": 642.3418746464796, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6631997440, "peak_system_memory_used_bytes": 53323919360, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 151.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 148.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 151.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 154.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 138.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 165.19, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.315099686034955}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-7440"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-09/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d6c84e227bb66da4c0cdd7cab54902164e489bc38b561066cbda2d965f3bc42e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-7440\"}}"}]}
{"case_id": "TC-10", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-10/seed-1731/checkpoint.json", "checkpoint_sha256": "24961cf285598755193bde9e22923bf78740bc9dc15bf9ea391cb7381b5bf38b", "completed_utc": "2026-08-02T03:07:48.180144+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 140, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.36339238355044, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 140, "prompt_tokens_per_second": 643.0179804581484, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6632452096, "peak_system_memory_used_bytes": 53585707008, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 159.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 158.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 154.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 154.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 156.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 164.51, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3016699630534276}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-1010"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 101, "path": "results/quants/UD-IQ2_M/cases/TC-10/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "7eb1a6b4bf40eb8a30b7f6e2ba07ce5fb3bd0bfa27484c234e636560c2d36455", "utf8": "{\n  \"kind\": \"tool_call\",\n  \"name\": \"get_invoice\",\n  \"arguments\": {\n    \"invoice_id\": \"INV-1010\"\n  }\n}"}]}
{"case_id": "TC-10", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-10/seed-2731/checkpoint.json", "checkpoint_sha256": "a4e1031b49372238d13b93ede6687112dfbb7003fac8722469f3428e4680cbbe", "completed_utc": "2026-08-02T03:07:53.573931+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 126, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.07058289088896, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 126, "prompt_tokens_per_second": 637.0116852071134, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6632509440, "peak_system_memory_used_bytes": 53845794816, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 169.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 144.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 163.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 149.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 139.46, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3260661700041965}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-1010"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-10/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "317f118bb6adea7882e5c29b2af7050ccc4ae34b5e6174d8a9a813feb2be56f7", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-1010\"}}"}]}
{"case_id": "TC-10", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-10/seed-731/checkpoint.json", "checkpoint_sha256": "4cf957760d2aa3d17951251718e55e6e23263ae9f0d55120d1d9a0017a6df673", "completed_utc": "2026-08-02T03:07:42.492726+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 94, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.35145189700411, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 94, "prompt_tokens_per_second": 643.2634641095457, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6632398848, "peak_system_memory_used_bytes": 53712662528, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 165.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 146.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 142.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 165.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 155.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 143.66, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3137343249982223}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-1010"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-10/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "317f118bb6adea7882e5c29b2af7050ccc4ae34b5e6174d8a9a813feb2be56f7", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-1010\"}}"}]}
{"case_id": "TC-11", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-11/seed-1731/checkpoint.json", "checkpoint_sha256": "28c1318024b9e5d4116a06aad821d8863ec986ac89e19915a9cf9adbf4ff51b8", "completed_utc": "2026-08-02T03:08:11.611600+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 212, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.87153931127657, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 212, "prompt_tokens_per_second": 645.7675607114089, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6632865792, "peak_system_memory_used_bytes": 55093874688, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 151.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 150.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 148.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 153.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 148.06, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 142.58, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3214306449517608}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-1111"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 101, "path": "results/quants/UD-IQ2_M/cases/TC-11/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1a7d7d2bf0318a76e9065d55f67a9149e735c6180b6220eab2b4a744013d49a0", "utf8": "{\n  \"kind\": \"tool_call\",\n  \"name\": \"get_invoice\",\n  \"arguments\": {\n    \"invoice_id\": \"INV-1111\"\n  }\n}"}]}
{"case_id": "TC-11", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-11/seed-2731/checkpoint.json", "checkpoint_sha256": "924db7aa95dc5abaf8550ef5d13827c5ddc97bdcd2ea8d6cbd1fbded011bbad5", "completed_utc": "2026-08-02T03:08:17.491833+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 146, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.84638424409085, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 146, "prompt_tokens_per_second": 645.5802818031203, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6633046016, "peak_system_memory_used_bytes": 54300372992, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 159.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 153.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 149.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 159.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 152.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 160.58, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3203446950064972}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-1111"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-11/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "bd32332a1d2ba332e0ccf2c95409f8527ca6deb74504b01f6ba3f6f4394f93f6", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-1111\"}}"}]}
{"case_id": "TC-11", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-11/seed-731/checkpoint.json", "checkpoint_sha256": "5d17f7d0c9ca4c41e599aa89cc483509a0f5c7792ec9be723e3712d73d9260f8", "completed_utc": "2026-08-02T03:08:03.760248+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 289, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.10552651144128, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 289, "prompt_tokens_per_second": 640.288122005104, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6632685568, "peak_system_memory_used_bytes": 53850808320, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 178.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 145.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 153.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 156.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 156.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 152.2, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3337036450393498}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"kind": "ask_user", "message": "Which amount should I refund on INV-1111: $9.00 or $19.00?"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 90, "path": "results/quants/UD-IQ2_M/cases/TC-11/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "718dc2b55c6b502c53c7ef28656330825490004d8b6575cd48c99b11e70c564d", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Which amount should I refund on INV-1111: $9.00 or $19.00?\"}"}]}
{"case_id": "TC-12", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-12/seed-1731/checkpoint.json", "checkpoint_sha256": "b5924a99fc734ad5f624f354ebc0f710bfaa771c73a3025d34202880da2f7aac", "completed_utc": "2026-08-02T03:08:25.457118+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 80, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.04697205381975, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 80, "prompt_tokens_per_second": 637.895452859603, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6633054208, "peak_system_memory_used_bytes": 51387080704, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 152.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 151.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 146.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 153.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 146.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 148.44, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3240189580246806}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-515", "note": "Stolen security key", "queue": "security"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 127, "path": "results/quants/UD-IQ2_M/cases/TC-12/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "530a64689607898b4b9710807e46435d2acefd9c3bc1b562285ddeae3b3976b5", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-515\",\"queue\":\"security\",\"note\":\"Stolen security key\"}}"}]}
{"case_id": "TC-12", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-12/seed-2731/checkpoint.json", "checkpoint_sha256": "1aaff3c233df2b1604a01fd09b58710e65a2dd787236f5f274a6e2da0f0d38e0", "completed_utc": "2026-08-02T03:08:30.370057+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 111, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.12145701695467, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 111, "prompt_tokens_per_second": 641.2066131683264, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6633222144, "peak_system_memory_used_bytes": 54383812608, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 174.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 140.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 157.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 158.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 157.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 158.16, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3042008760385215}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-515", "note": "Stolen security key.", "queue": "security"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 128, "path": "results/quants/UD-IQ2_M/cases/TC-12/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "61f7c1032c41cca6ad19c9628767cbd10bcc53cda75048231b8ea84627011f3e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-515\",\"queue\":\"security\",\"note\":\"Stolen security key.\"}}"}]}
{"case_id": "TC-12", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-12/seed-731/checkpoint.json", "checkpoint_sha256": "4978de773d365555f6f9b2135320df9dd56049c25383874fdc0fe72c130fe210", "completed_utc": "2026-08-02T03:08:21.440349+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 78, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.81430570560052, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 78, "prompt_tokens_per_second": 638.8230244455764, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6633046016, "peak_system_memory_used_bytes": 54208815104, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 156.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 142.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 155.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 162.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 139.36, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3234968879260123}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-515", "note": "Stolen security key", "queue": "security"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 127, "path": "results/quants/UD-IQ2_M/cases/TC-12/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "530a64689607898b4b9710807e46435d2acefd9c3bc1b562285ddeae3b3976b5", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-515\",\"queue\":\"security\",\"note\":\"Stolen security key\"}}"}]}
{"case_id": "TC-13", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-13/seed-1731/checkpoint.json", "checkpoint_sha256": "acdb0e9404d3d6abf5767078ccb0b077d493c0220c2ef1366e1cc6bda2d2fdd5", "completed_utc": "2026-08-02T03:08:38.488049+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 83, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.344410716118205, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 83, "prompt_tokens_per_second": 638.9510091258873, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6633582592, "peak_system_memory_used_bytes": 54144843776, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 148.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 151.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 139.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 158.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 151.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 140.28, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3216741499491036}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-616", "note": "Repeated invoice mismatch.", "queue": "billing"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 133, "path": "results/quants/UD-IQ2_M/cases/TC-13/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "7c1e6c5631edd80383226c77059e81122ffb16a425f5b71361e5ec9e81927b4f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-616\",\"queue\":\"billing\",\"note\":\"Repeated invoice mismatch.\"}}"}]}
{"case_id": "TC-13", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-13/seed-2731/checkpoint.json", "checkpoint_sha256": "aeeb6da15843b4a5905b56d163092cd3fcd21532dab9f4afcee995bb328d8a78", "completed_utc": "2026-08-02T03:08:42.814008+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 93, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.923480301036236, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 93, "prompt_tokens_per_second": 641.8012601938271, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6633758720, "peak_system_memory_used_bytes": 50614001664, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 163.36, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 160.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 139.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 167.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 158.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 143.62, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3142082260455936}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-616", "note": "Repeated invoice mismatch.", "queue": "billing"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 133, "path": "results/quants/UD-IQ2_M/cases/TC-13/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "7c1e6c5631edd80383226c77059e81122ffb16a425f5b71361e5ec9e81927b4f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-616\",\"queue\":\"billing\",\"note\":\"Repeated invoice mismatch.\"}}"}]}
{"case_id": "TC-13", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-13/seed-731/checkpoint.json", "checkpoint_sha256": "261e8636d4ef139bc11063c7eb1bc75f8ce9416fb8456298106f683788358b48", "completed_utc": "2026-08-02T03:08:34.444689+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 80, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.25823428778202, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 80, "prompt_tokens_per_second": 638.2280077265671, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6633406464, "peak_system_memory_used_bytes": 54238158848, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 159.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 153.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 147.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 159.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 149.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 152.9, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.322489155922085}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-616", "note": "Repeated invoice mismatch", "queue": "billing"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 132, "path": "results/quants/UD-IQ2_M/cases/TC-13/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "60c790db56251d37104cdee67d0e98f5f6abdec0e6c07276196b272bb73c0c87", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-616\",\"queue\":\"billing\",\"note\":\"Repeated invoice mismatch\"}}"}]}
{"case_id": "TC-14", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-14/seed-1731/checkpoint.json", "checkpoint_sha256": "84426346cc42fc90e1e3d197c41cb3ba248e06203d1f054fb445bc65043b11f6", "completed_utc": "2026-08-02T03:09:04.632984+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 114, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.034351408260854, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 114, "prompt_tokens_per_second": 675.4327905145557, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6634369024, "peak_system_memory_used_bytes": 50287292416, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 178.06, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 157.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 156.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 169.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 155.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 152.86, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.436343741021119}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "owner@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-714", "note": "Estate access review", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 89, "path": "results/quants/UD-IQ2_M/cases/TC-14/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d69329385f42792b3f62ce14173757205b66eb3a684ab367f8fd6912e0cea5e6", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"owner@example.test\"}}"}, {"bytes": 125, "path": "results/quants/UD-IQ2_M/cases/TC-14/seed-1731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "6a11d3f7a4a318fccc827137a8c22d0f8140c674678aa4eb3206a8e766bdebec", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-714\",\"queue\":\"legal\",\"note\":\"Estate access review\"}}"}]}
{"case_id": "TC-14", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-14/seed-2731/checkpoint.json", "checkpoint_sha256": "f5c4af90b4ba36b8bd50d60c93ef884129996df54651fb213bcf2d3bece8407c", "completed_utc": "2026-08-02T03:09:11.989846+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 85, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.60802290722842, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 85, "prompt_tokens_per_second": 677.1051929749799, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6634590208, "peak_system_memory_used_bytes": 49943457792, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 152.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 150.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 152.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 167.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 153.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 170.88, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.430610114010051}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "owner@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-714", "note": "Estate access review", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 89, "path": "results/quants/UD-IQ2_M/cases/TC-14/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d69329385f42792b3f62ce14173757205b66eb3a684ab367f8fd6912e0cea5e6", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"owner@example.test\"}}"}, {"bytes": 125, "path": "results/quants/UD-IQ2_M/cases/TC-14/seed-2731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "6a11d3f7a4a318fccc827137a8c22d0f8140c674678aa4eb3206a8e766bdebec", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-714\",\"queue\":\"legal\",\"note\":\"Estate access review\"}}"}]}
{"case_id": "TC-14", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-14/seed-731/checkpoint.json", "checkpoint_sha256": "23461723936ecf2cf5b9182332aaf050f35eb85484895c5c296f9d77627c382f", "completed_utc": "2026-08-02T03:08:54.771565+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 158, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.94591384655974, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 158, "prompt_tokens_per_second": 688.4698598487203, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6634139648, "peak_system_memory_used_bytes": 50181308416, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 173.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 153.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 158.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 161.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 167.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 160.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4473088619997725}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "owner@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-714", "note": "Estate access review", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 111, "path": "results/quants/UD-IQ2_M/cases/TC-14/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d1acaa4fbab76983a0ca321e1b933539b0c089a976ef156fda0f25a47d0a6bdc", "utf8": "{\n  \"kind\": \"tool_call\",\n  \"name\": \"search_customers\",\n  \"arguments\": {\n    \"email\": \"owner@example.test\"\n  }\n}"}, {"bytes": 125, "path": "results/quants/UD-IQ2_M/cases/TC-14/seed-731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "6a11d3f7a4a318fccc827137a8c22d0f8140c674678aa4eb3206a8e766bdebec", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-714\",\"queue\":\"legal\",\"note\":\"Estate access review\"}}"}]}
{"case_id": "TC-15", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-15/seed-1731/checkpoint.json", "checkpoint_sha256": "82f9c4d413c9eea208f616ce0da61bd74828d4ff50a4915119de7122b116b6a3", "completed_utc": "2026-08-02T03:09:18.513186+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 45, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 35.00180453747838, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 45, "prompt_tokens_per_second": 639.2944683876682, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6634954752, "peak_system_memory_used_bytes": 49916719104, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 152.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 151.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 139.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 154.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 148.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 152.1, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 2, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3058645849814638}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8080"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-15/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ad3a40d83ab15f4a3e7b92cb50eae47c0b668ecd18b6c8c66f8c6d0c83a85d7c", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-8080\"}}"}]}
{"case_id": "TC-15", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-15/seed-2731/checkpoint.json", "checkpoint_sha256": "2e08067fa9444e283614c468965639acbf4e53a7171813391257bdf020d5cbe7", "completed_utc": "2026-08-02T03:09:22.787040+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 92, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.98479475954465, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 92, "prompt_tokens_per_second": 639.6565200501116, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6635139072, "peak_system_memory_used_bytes": 49916526592, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 175.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 141.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 152.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 160.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 156.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 142.63, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3073639310896397}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8080"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-15/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ad3a40d83ab15f4a3e7b92cb50eae47c0b668ecd18b6c8c66f8c6d0c83a85d7c", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-8080\"}}"}]}
{"case_id": "TC-15", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-15/seed-731/checkpoint.json", "checkpoint_sha256": "6e0e9da151fddd71aa3664f77a11b2c6a48960f9aa98b6696388a76898702888", "completed_utc": "2026-08-02T03:09:15.676232+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 74, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.795365633463746, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 74, "prompt_tokens_per_second": 639.9240831526718, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6634774528, "peak_system_memory_used_bytes": 49947246592, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 165.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 141.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 144.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 166.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 151.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 138.59, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3047892710892484}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8080"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-15/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ad3a40d83ab15f4a3e7b92cb50eae47c0b668ecd18b6c8c66f8c6d0c83a85d7c", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-8080\"}}"}]}
{"case_id": "TC-16", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-16/seed-1731/checkpoint.json", "checkpoint_sha256": "1bfc1051d00fc6a891f6d9979e562954d79b63a3492096951118402d7d1bad96", "completed_utc": "2026-08-02T03:09:31.282864+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 106, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.3326509988858, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 106, "prompt_tokens_per_second": 637.4739075069297, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6635139072, "peak_system_memory_used_bytes": 49644367872, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 152.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 153.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 140.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 151.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 139.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 155.38, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.322019652929157}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"kind": "ask_user", "message": "Please provide the customer ID or email for the account to escalate."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 100, "path": "results/quants/UD-IQ2_M/cases/TC-16/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "4db77f3822590c338d9aceac65e5ef0cded1cb3629ec32478a243bce3e05a1ab", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Please provide the customer ID or email for the account to escalate.\"}"}]}
{"case_id": "TC-16", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-16/seed-2731/checkpoint.json", "checkpoint_sha256": "b7cbc85446e6b0faf5e15d511a5139f1ec7e04b5b1b253332b1df60cab6dd6dc", "completed_utc": "2026-08-02T03:09:35.291377+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 83, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.70454353490004, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 83, "prompt_tokens_per_second": 641.799708823155, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6635257856, "peak_system_memory_used_bytes": 53675008000, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 155.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 151.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 141.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 155.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 162.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 146.49, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3121513799997047}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"kind": "ask_user", "message": "Please provide the customer ID (e.g., C-123) or email address for the account to escalate."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 122, "path": "results/quants/UD-IQ2_M/cases/TC-16/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dfaaa49b8c46f21b4a9d62b3ee40b3c119e29f4651b3c4c8fb4df1ab6fcd08a8", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Please provide the customer ID (e.g., C-123) or email address for the account to escalate.\"}"}]}
{"case_id": "TC-16", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-16/seed-731/checkpoint.json", "checkpoint_sha256": "779f3e6997072b9de9309e45fcad08e20758eb87e2daccdca33e8c0897404c9c", "completed_utc": "2026-08-02T03:09:26.640608+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 75, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.278427671375, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 75, "prompt_tokens_per_second": 643.0908173318808, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6635139072, "peak_system_memory_used_bytes": 49795035136, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 173.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 159.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 153.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 162.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 158.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 144.39, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3097270809812471}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"kind": "ask_user", "message": "Please provide the customer ID (e.g., C-123) or email address for the account to escalate."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 122, "path": "results/quants/UD-IQ2_M/cases/TC-16/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dfaaa49b8c46f21b4a9d62b3ee40b3c119e29f4651b3c4c8fb4df1ab6fcd08a8", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Please provide the customer ID (e.g., C-123) or email address for the account to escalate.\"}"}]}
{"case_id": "TC-17", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-17/seed-1731/checkpoint.json", "checkpoint_sha256": "7f4ff91b2e34d2a6d39be202a01a7c583cb89b8826794e19c3123ab0c6fb7f09", "completed_utc": "2026-08-02T03:09:45.187787+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 87, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.28560619316274, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 87, "prompt_tokens_per_second": 638.8897687249938, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6635520000, "peak_system_memory_used_bytes": 52529131520, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 152.06, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 149.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 150.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 155.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 143.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 160.98, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3047279559541494}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"amount_cents": 325, "invoice_id": "INV-3250", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 120, "path": "results/quants/UD-IQ2_M/cases/TC-17/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "68524c26479d8c763f3370fa7c6d957caa35a3984bbee1ad0b5eac104ca49eea", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-3250\",\"amount_cents\":325,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-17", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-17/seed-2731/checkpoint.json", "checkpoint_sha256": "f1e95991e2f65d3669efe279a093ce18407e114c6d463fbe5fcb8a25e46e1cd5", "completed_utc": "2026-08-02T03:09:49.351972+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 84, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.338427275503335, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 84, "prompt_tokens_per_second": 636.9274680554304, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6635642880, "peak_system_memory_used_bytes": 53003210752, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 151.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 152.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 148.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 150.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 149.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 156.03, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3225573670351878}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"amount_cents": 325, "invoice_id": "INV-3250", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 120, "path": "results/quants/UD-IQ2_M/cases/TC-17/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "68524c26479d8c763f3370fa7c6d957caa35a3984bbee1ad0b5eac104ca49eea", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-3250\",\"amount_cents\":325,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-17", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-17/seed-731/checkpoint.json", "checkpoint_sha256": "a36300171afff1daa94f206956a43c18840b589618f15d82cb15c940db9bade9", "completed_utc": "2026-08-02T03:09:40.939327+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 138, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.33616498875615, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 138, "prompt_tokens_per_second": 638.3113384779015, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6635384832, "peak_system_memory_used_bytes": 56259399680, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 150.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 148.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 152.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 156.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 163.19, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3045250369468704}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-3250"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-17/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "eb276a823cf317bcb00410a85266c101c0e248ef9384e4bbf964c4a675ae2dde", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-3250\"}}"}]}
{"case_id": "TC-18", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-18/seed-1731/checkpoint.json", "checkpoint_sha256": "4ad5c697e823f40940228bab293aadc1bf659ef91df5d80cd72b41cd0736c9c6", "completed_utc": "2026-08-02T03:10:21.994510+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 206, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.07219269115079, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 206, "prompt_tokens_per_second": 673.5019954287843, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6636134400, "peak_system_memory_used_bytes": 51084513280, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 153.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 157.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 158.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 164.06, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 166.46, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4263516620267183}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "zoë@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-918"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 88, "path": "results/quants/UD-IQ2_M/cases/TC-18/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcd2013a41e2c21a2375fd2fe8398c8d7a72305abece93dd5f039d3501fb6e9f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"zoë@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ2_M/cases/TC-18/seed-1731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "08453e209245616b2e7178a214ff1e085fab00d101b565907129428a4c41c828", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-918\"}}"}]}
{"case_id": "TC-18", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-18/seed-2731/checkpoint.json", "checkpoint_sha256": "bb352c53d495a7baf800cb2214d33f546cf74f451c142135534a243c839b2323", "completed_utc": "2026-08-02T03:10:37.063950+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 321, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.95732294429666, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 321, "prompt_tokens_per_second": 673.905656075833, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6636294144, "peak_system_memory_used_bytes": 50751692800, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 165.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 149.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 151.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 160.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 155.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 159.64, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 10, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4276229600654915}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "zoë@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-918"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 88, "path": "results/quants/UD-IQ2_M/cases/TC-18/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcd2013a41e2c21a2375fd2fe8398c8d7a72305abece93dd5f039d3501fb6e9f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"zoë@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ2_M/cases/TC-18/seed-2731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "08453e209245616b2e7178a214ff1e085fab00d101b565907129428a4c41c828", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-918\"}}"}]}
{"case_id": "TC-18", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-18/seed-731/checkpoint.json", "checkpoint_sha256": "ed185f73f3614ed4353daee8da9232d0e66ec201f11b7e04120e03ed3f3bb4c2", "completed_utc": "2026-08-02T03:10:01.285165+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 115, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.04551619071528, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 115, "prompt_tokens_per_second": 672.071840189896, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6635958272, "peak_system_memory_used_bytes": 51719888896, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 191.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 150.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 150.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 169.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 150.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 149.99, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.433466529008001}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"email": "zoë@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-918"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 88, "path": "results/quants/UD-IQ2_M/cases/TC-18/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcd2013a41e2c21a2375fd2fe8398c8d7a72305abece93dd5f039d3501fb6e9f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"zoë@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ2_M/cases/TC-18/seed-731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "08453e209245616b2e7178a214ff1e085fab00d101b565907129428a4c41c828", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-918\"}}"}]}
{"case_id": "TC-19", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-19/seed-1731/checkpoint.json", "checkpoint_sha256": "7406ee792b496fae0fe28a8960b4f37ba418b3f217573f48eeb20da2a5114273", "completed_utc": "2026-08-02T03:10:59.314779+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 97, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.335918556617095, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 97, "prompt_tokens_per_second": 645.151320821776, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6636556288, "peak_system_memory_used_bytes": 50832297984, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 153.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 153.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 140.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 154.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 147.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 149.67, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3182040009414777}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"kind": "ask_user", "message": "Please specify whether this should be escalated to the security, billing, or legal queue for customer C-919."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 140, "path": "results/quants/UD-IQ2_M/cases/TC-19/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a2b767aa601b19d8df22bfe3dc1da591038814721145c5f85a7eb9b916208f3d", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Please specify whether this should be escalated to the security, billing, or legal queue for customer C-919.\"}"}]}
{"case_id": "TC-19", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-19/seed-2731/checkpoint.json", "checkpoint_sha256": "c4a88a856690a35ddc8b256c98505c9e8fd7686cd2fa32f039f4b3efd6399a56", "completed_utc": "2026-08-02T03:11:03.669221+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 94, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.93667200387023, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 94, "prompt_tokens_per_second": 645.6776328762039, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6636687360, "peak_system_memory_used_bytes": 50760495104, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 151.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 149.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 149.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 155.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 135.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 148.48, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3172973711043596}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"customer_id": "C-919"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 78, "path": "results/quants/UD-IQ2_M/cases/TC-19/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "7ac566880f0406939dcaec8ba1568d862825ae3e1b27a7b9ca4e466830f07f14", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-919\"}}"}]}
{"case_id": "TC-19", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-19/seed-731/checkpoint.json", "checkpoint_sha256": "fc1fc879b5a33f5e9c4f8c41dd4d6ad064c8f234c56c06b6eceba6893d84d344", "completed_utc": "2026-08-02T03:10:54.839375+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 547, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.93951878477909, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 547, "prompt_tokens_per_second": 640.7103521144593, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6636425216, "peak_system_memory_used_bytes": 50674991104, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 151.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 151.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 150.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 148.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 146.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 143.58, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 15, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.328330593998544}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"kind": "ask_user", "message": "Which queue should C-919 be escalated to: security, billing, or legal?"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 102, "path": "results/quants/UD-IQ2_M/cases/TC-19/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "57861fd8b4c51433e3c03edbbb540cfb34001e800fb15c96d9f43abdbbf001ac", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Which queue should C-919 be escalated to: security, billing, or legal?\"}"}]}
{"case_id": "TC-20", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-20/seed-1731/checkpoint.json", "checkpoint_sha256": "d035f04b491a210abac1edecfdb4fb17c0a738d6da3755ba6caf74f8855367df", "completed_utc": "2026-08-02T03:11:21.773534+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 77, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.87546289250748, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 77, "prompt_tokens_per_second": 672.6962312463443, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6636830720, "peak_system_memory_used_bytes": 50409259008, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 176.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 154.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 149.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 152.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 153.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 151.65, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4206531969830394}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8820"}, "kind": "tool_call", "name": "get_invoice"}, {"arguments": {"amount_cents": 1275, "invoice_id": "INV-8820", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ2_M/cases/TC-20/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "18ba214e3084f670760bb3f93bae86f57a2254f07c8f8f9539bebe37154d9e29", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-8820\"}}"}, {"bytes": 121, "path": "results/quants/UD-IQ2_M/cases/TC-20/seed-1731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "8fcbfee0badae01da9ff392b97f3fc374979123ed9ef3185e5c3e5ed6d90529f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-8820\",\"amount_cents\":1275,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-20", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-20/seed-2731/checkpoint.json", "checkpoint_sha256": "735e6ab39577caf1b150d699f9e88e482594447ecaf12daf7d6db9a652c1cd06", "completed_utc": "2026-08-02T03:11:30.368223+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 88, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.6140488389211, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 88, "prompt_tokens_per_second": 684.2754496464221, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6637031424, "peak_system_memory_used_bytes": 50393526272, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 161.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 150.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 156.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 162.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 162.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 169.79, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4388442490017042}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8820"}, "kind": "tool_call", "name": "get_invoice"}, {"arguments": {"amount_cents": 1275, "invoice_id": "INV-8820", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 101, "path": "results/quants/UD-IQ2_M/cases/TC-20/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "47574e125f79fd186dad5f31fb6c00b5a205cd234c3d15f5808c82309e052810", "utf8": "{\n  \"kind\": \"tool_call\",\n  \"name\": \"get_invoice\",\n  \"arguments\": {\n    \"invoice_id\": \"INV-8820\"\n  }\n}"}, {"bytes": 155, "path": "results/quants/UD-IQ2_M/cases/TC-20/seed-2731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "7ca46c288a7e360115f19b7c80fadb914fe735c61fcc1e6d92c6c69d56a525e3", "utf8": "{\n  \"kind\": \"tool_call\",\n  \"name\": \"issue_refund\",\n  \"arguments\": {\n    \"invoice_id\": \"INV-8820\",\n    \"amount_cents\": 1275,\n    \"reason\": \"duplicate\"\n  }\n}"}]}
{"case_id": "TC-20", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TC-20/seed-731/checkpoint.json", "checkpoint_sha256": "74ad5724ac9891af3c79b1e5c347d1fd66609b6565fb52bd91ce07c1d15ad277", "completed_utc": "2026-08-02T03:11:14.251686+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 161, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.05966449326518, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 161, "prompt_tokens_per_second": 683.6366630863888, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6636789760, "peak_system_memory_used_bytes": 50343096320, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 176.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 158.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 149.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 149.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 151.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 140.74, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4401222030865029}, "quant": "UD-IQ2_M", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8820"}, "kind": "tool_call", "name": "get_invoice"}, {"arguments": {"amount_cents": 1275, "invoice_id": "INV-8820", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 101, "path": "results/quants/UD-IQ2_M/cases/TC-20/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "47574e125f79fd186dad5f31fb6c00b5a205cd234c3d15f5808c82309e052810", "utf8": "{\n  \"kind\": \"tool_call\",\n  \"name\": \"get_invoice\",\n  \"arguments\": {\n    \"invoice_id\": \"INV-8820\"\n  }\n}"}, {"bytes": 155, "path": "results/quants/UD-IQ2_M/cases/TC-20/seed-731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "7ca46c288a7e360115f19b7c80fadb914fe735c61fcc1e6d92c6c69d56a525e3", "utf8": "{\n  \"kind\": \"tool_call\",\n  \"name\": \"issue_refund\",\n  \"arguments\": {\n    \"invoice_id\": \"INV-8820\",\n    \"amount_cents\": 1275,\n    \"reason\": \"duplicate\"\n  }\n}"}]}
{"case_id": "TS-PATCH-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TS-PATCH-01/seed-1731/checkpoint.json", "checkpoint_sha256": "373e4440c0165737cfea284ff1f38cfe49888391f0b14f39f62a07d6aab0cbc9", "completed_utc": "2026-08-02T03:15:42.812103+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.86962683817229, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 614.3492220301308, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6749671424, "peak_system_memory_used_bytes": 51908976640, "per_gpu": {"0": {"peak_memory_used_mib": 17904.0, "peak_power_draw_w": 155.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16852.0, "peak_power_draw_w": 152.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16822.0, "peak_power_draw_w": 142.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19392.0, "peak_power_draw_w": 161.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16852.0, "peak_power_draw_w": 150.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 152.23, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 114, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3707620210479945}, "quant": "UD-IQ2_M", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "typescript_patch"}, "seed": 1731, "status": "completed", "task": "typescript_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/TS-PATCH-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "TS-PATCH-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TS-PATCH-01/seed-2731/checkpoint.json", "checkpoint_sha256": "e8d834511b1d408347ac9cfd3ca42e32703bb97d19f95006ad9fbbf3c76e8b88", "completed_utc": "2026-08-02T03:17:49.070228+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.90041971489537, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 614.2373425438069, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6819237888, "peak_system_memory_used_bytes": 50955317248, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 164.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 148.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 150.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 151.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 146.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 170.02, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 113, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.372462696977891}, "quant": "UD-IQ2_M", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "typescript_patch"}, "seed": 2731, "status": "completed", "task": "typescript_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/TS-PATCH-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "TS-PATCH-01", "checkpoint_path": "results/quants/UD-IQ2_M/cases/TS-PATCH-01/seed-731/checkpoint.json", "checkpoint_sha256": "bac51e7db706caa32f03b4f301678bd34a3a0f3c6df41f16041c16dbe519786b", "completed_utc": "2026-08-02T03:13:36.542608+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.90916848550215, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 613.2289508777417, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6680125440, "peak_system_memory_used_bytes": 50700128256, "per_gpu": {"0": {"peak_memory_used_mib": 17637.0, "peak_power_draw_w": 168.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 142.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 16555.0, "peak_power_draw_w": 149.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 19125.0, "peak_power_draw_w": 149.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 16585.0, "peak_power_draw_w": 148.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 15165.0, "peak_power_draw_w": 166.65, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 113, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3627956029959023}, "quant": "UD-IQ2_M", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "typescript_patch"}, "seed": 731, "status": "completed", "task": "typescript_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ2_M/cases/TS-PATCH-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}

QUANT UD-IQ3_XXS
manifest: work/runpod-live/full-run-jsqxbdwx8mi8ti/pulled/evidence/quants/UD-IQ3_XXS/base/evidence-manifest.json
manifest_sha256: 5e9b16ec8ee0e9eec4425a90e112b4295b68e1cf601fdc2f3e821657fdcc52f3
completed_utc: 2026-08-02T01:29:30.719595+00:00
executed_rows: 177
decision_included_rows: 159
multilingual_diagnostic_rows: 18
decision_score: 320/426
decision_percent: 75.117371
task_summary_json:
{"multilingual": {"included": 18.0, "max_score": 180.0, "score": 111.0}, "python_patch": {"included": 3.0, "max_score": 3.0, "score": 0.0}, "reasoning": {"included": 90.0, "max_score": 90.0, "score": 66.0}, "strict_json": {"included": 3.0, "max_score": 30.0, "score": 20.0}, "tool_calling": {"included": 60.0, "max_score": 300.0, "score": 234.0}, "typescript_patch": {"included": 3.0, "max_score": 3.0, "score": 0.0}}
performance_summary_json:
{"decode_tokens_per_second": {"maximum": 35.55486026051046, "median": 34.16731365541371, "minimum": 31.95090562758247}, "peak_process_rss_bytes": {"maximum": 6992371712.0, "median": 5609381888.0, "minimum": 2034728960.0}, "peak_system_memory_used_bytes": {"maximum": 66999607296.0, "median": 59786915840.0, "minimum": 50813358080.0}, "prompt_tokens_per_second": {"maximum": 715.6704703729246, "median": 308.8006904731757, "minimum": 222.19989215521156}, "swap_delta_bytes": {"maximum": 0.0, "median": 0.0, "minimum": 0.0}, "time_to_first_token_seconds": {"maximum": 1.9395060440292582, "median": 0.8263086740626022, "minimum": 0.5600969530642033}}

ROW-LEVEL RECORDS
{"case_id": "LANG-de", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-de/seed-1731/checkpoint.json", "checkpoint_sha256": "706d9570dd2eceb6dcce288fa0be3a43a28aed2d8f711eb972aa13938a40c9b5", "completed_utc": "2026-08-02T00:22:07.221503+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 307, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.6635009593999, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 307, "prompt_tokens_per_second": 287.5015565370896, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2078781440, "peak_system_memory_used_bytes": 55538401280, "per_gpu": {"0": {"peak_memory_used_mib": 19783.0, "peak_power_draw_w": 144.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18217.0, "peak_power_draw_w": 143.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17803.0, "peak_power_draw_w": 141.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21981.0, "peak_power_draw_w": 146.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19369.0, "peak_power_draw_w": 145.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16799.0, "peak_power_draw_w": 156.3, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6097194700269029}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 1.0, "hold_status": 0.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "de", "max_score": 10.0, "passed": false, "score": 4.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 165, "path": "results/quants/UD-IQ3_XXS/cases/LANG-de/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1c9c57c373be5dc5bbc4945122545fb894546e751bcd40832e6608452cd0d6a0", "utf8": "Mara Chen hat den Versand VX-44 nicht genehmigt. Er bleibt bis zum 17. August 2026 zurückgehalten. Das genehmigte Limit beträgt 12.450,75 €, nicht 12.540,75 €."}]}
{"case_id": "LANG-de", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-de/seed-2731/checkpoint.json", "checkpoint_sha256": "e4dac51f00b9a53ad0b27ad6687dcad0a0718a7c3c821c0769af89d4bad4bf14", "completed_utc": "2026-08-02T00:22:16.022824+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 270, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.577017424895836, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 270, "prompt_tokens_per_second": 283.24829870607846, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2117951488, "peak_system_memory_used_bytes": 55428554752, "per_gpu": {"0": {"peak_memory_used_mib": 19783.0, "peak_power_draw_w": 144.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18217.0, "peak_power_draw_w": 146.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17803.0, "peak_power_draw_w": 141.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21981.0, "peak_power_draw_w": 146.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19369.0, "peak_power_draw_w": 143.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16799.0, "peak_power_draw_w": 141.59, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6067349959630519}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 1.0, "hold_status": 0.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "de", "max_score": 10.0, "passed": false, "score": 4.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 160, "path": "results/quants/UD-IQ3_XXS/cases/LANG-de/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "4f1aad40de1d878ae9da9ade1912e8db6652272bf91f47d26a583451daa009c6", "utf8": "Mara Chen hat den Versand VX-44 nicht genehmigt. Er bleibt bis zum 17. August 2026 angehalten. Das genehmigte Limit beträgt 12.450,75 €, nicht 12.540,75 €."}]}
{"case_id": "LANG-de", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-de/seed-731/checkpoint.json", "checkpoint_sha256": "09c6a7b6dbd04d2aa9e1decfc30881a70295ca079f303ea35cf5d156987de920", "completed_utc": "2026-08-02T00:21:57.346769+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 1024, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.05980162759951, "finish_reason": "length", "output_token_cap": 1024, "output_tokens": 1024, "prompt_tokens_per_second": 222.19989215521156, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2034728960, "peak_system_memory_used_bytes": 55522029568, "per_gpu": {"0": {"peak_memory_used_mib": 19783.0, "peak_power_draw_w": 145.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18217.0, "peak_power_draw_w": 142.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17803.0, "peak_power_draw_w": 138.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21981.0, "peak_power_draw_w": 143.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19369.0, "peak_power_draw_w": 150.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16799.0, "peak_power_draw_w": 143.66, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 28, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7332048380048946}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 0.0, "hold_status": 0.0, "name": 0.0, "negation": 0.0, "semantic_adequacy": 0.0, "shipment_identifier": 0.0}, "language": "de", "max_score": 10.0, "passed": false, "score": 0.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/LANG-de/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "LANG-en", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-en/seed-1731/checkpoint.json", "checkpoint_sha256": "bec1b4356b6e7d0f158a5e4b4ebcabc20b594fc1dee5b761780ac0f8d2857ce3", "completed_utc": "2026-08-02T00:22:22.965802+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 60, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 35.55486026051046, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 60, "prompt_tokens_per_second": 290.09508876358905, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2157166592, "peak_system_memory_used_bytes": 55402938368, "per_gpu": {"0": {"peak_memory_used_mib": 19783.0, "peak_power_draw_w": 138.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18217.0, "peak_power_draw_w": 138.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17803.0, "peak_power_draw_w": 139.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21981.0, "peak_power_draw_w": 138.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19369.0, "peak_power_draw_w": 137.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16799.0, "peak_power_draw_w": 142.69, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 2, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5609198809834197}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": true, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 2.0, "semantic_adequacy": 2.0, "shipment_identifier": 1.0}, "language": "en", "max_score": 10.0, "passed": true, "score": 10.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 136, "path": "results/quants/UD-IQ3_XXS/cases/LANG-en/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "c3206d9f219883f5fee0a386cc815d5ca321881c6b6b0336ef4a9e9f98a1b9aa", "utf8": "Mara Chen did not approve shipment VX-44. It remains on hold until 17 August 2026. The approved limit is €12,450.75, not €12,540.75."}]}
{"case_id": "LANG-en", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-en/seed-2731/checkpoint.json", "checkpoint_sha256": "ef49e2ec50e3a58857db39be43a851238dbacc286ba703f6d9c70ad86ce97d6b", "completed_utc": "2026-08-02T00:22:28.141341+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 148, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.84619993949005, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 148, "prompt_tokens_per_second": 290.31416402382774, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2195054592, "peak_system_memory_used_bytes": 55438249984, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 154.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 141.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 139.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 143.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 137.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 152.73, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5600969530642033}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": true, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 2.0, "semantic_adequacy": 2.0, "shipment_identifier": 1.0}, "language": "en", "max_score": 10.0, "passed": true, "score": 10.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 136, "path": "results/quants/UD-IQ3_XXS/cases/LANG-en/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "c3206d9f219883f5fee0a386cc815d5ca321881c6b6b0336ef4a9e9f98a1b9aa", "utf8": "Mara Chen did not approve shipment VX-44. It remains on hold until 17 August 2026. The approved limit is €12,450.75, not €12,540.75."}]}
{"case_id": "LANG-en", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-en/seed-731/checkpoint.json", "checkpoint_sha256": "35e9f848c14bdf903016b35d1d5eac8e595450c5f944c23aa7cd81af21bf0a65", "completed_utc": "2026-08-02T00:22:20.318778+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 115, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.91455800446603, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 115, "prompt_tokens_per_second": 291.3210302439169, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2156867584, "peak_system_memory_used_bytes": 55431155712, "per_gpu": {"0": {"peak_memory_used_mib": 19783.0, "peak_power_draw_w": 142.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18217.0, "peak_power_draw_w": 141.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17803.0, "peak_power_draw_w": 138.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21981.0, "peak_power_draw_w": 141.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19369.0, "peak_power_draw_w": 138.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16799.0, "peak_power_draw_w": 144.38, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5975213720230386}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": true, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 2.0, "semantic_adequacy": 2.0, "shipment_identifier": 1.0}, "language": "en", "max_score": 10.0, "passed": true, "score": 10.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 136, "path": "results/quants/UD-IQ3_XXS/cases/LANG-en/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "c3206d9f219883f5fee0a386cc815d5ca321881c6b6b0336ef4a9e9f98a1b9aa", "utf8": "Mara Chen did not approve shipment VX-44. It remains on hold until 17 August 2026. The approved limit is €12,450.75, not €12,540.75."}]}
{"case_id": "LANG-es", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-es/seed-1731/checkpoint.json", "checkpoint_sha256": "9b6e8e03e1160904b79b8d7ffe7a9e5298b752ca478d6c316a6e9407c75492e1", "completed_utc": "2026-08-02T00:22:59.957946+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 551, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.531490400433675, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 551, "prompt_tokens_per_second": 289.57867225412093, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2273746944, "peak_system_memory_used_bytes": 55639912448, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 145.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 143.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 139.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 142.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 139.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 143.13, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5980000109411776}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "es", "max_score": 10.0, "passed": false, "score": 7.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 138, "path": "results/quants/UD-IQ3_XXS/cases/LANG-es/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1f04cb6fb157957d61852adc6bf44bdcab8fc3350bde213ec6f46b73c77f989c", "utf8": "Mara Chen no aprobó el envío VX-44. Sigue en espera hasta el 17 de agosto de 2026. El límite aprobado es €12,450.75, no €12,540.75."}]}
{"case_id": "LANG-es", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-es/seed-2731/checkpoint.json", "checkpoint_sha256": "84b533e54d46432dbd7302d435ddb7a3b63c59dacd7a3ce89466fb2da3cf3a00", "completed_utc": "2026-08-02T00:23:10.335579+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 326, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.69613581898153, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 326, "prompt_tokens_per_second": 289.3715844476597, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2314620928, "peak_system_memory_used_bytes": 55666778112, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 146.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 143.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 139.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 144.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 142.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 143.86, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5953074930002913}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "es", "max_score": 10.0, "passed": false, "score": 7.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 142, "path": "results/quants/UD-IQ3_XXS/cases/LANG-es/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dbb204af3a0902e8046eb17d2711627ef6edaa818702587394eeec1640a89069", "utf8": "Mara Chen no aprobó el envío VX-44. Permanece en espera hasta el 17 de agosto de 2026. El límite aprobado es €12,450.75, no €12,540.75."}]}
{"case_id": "LANG-es", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-es/seed-731/checkpoint.json", "checkpoint_sha256": "57094acc0898d8433ae2eee6e3c18359da85db0cf6279d84dafd713716c9ec98", "completed_utc": "2026-08-02T00:22:43.013168+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 479, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.592517833381784, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 479, "prompt_tokens_per_second": 288.94659641154095, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2233286656, "peak_system_memory_used_bytes": 55519023104, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 145.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 140.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 139.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 141.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 141.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 160.64, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5970373810268939}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "es", "max_score": 10.0, "passed": false, "score": 7.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 142, "path": "results/quants/UD-IQ3_XXS/cases/LANG-es/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dbb204af3a0902e8046eb17d2711627ef6edaa818702587394eeec1640a89069", "utf8": "Mara Chen no aprobó el envío VX-44. Permanece en espera hasta el 17 de agosto de 2026. El límite aprobado es €12,450.75, no €12,540.75."}]}
{"case_id": "LANG-fr", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-fr/seed-1731/checkpoint.json", "checkpoint_sha256": "b998486490d13530e2fa9c0182cdecfcb5516a15c7e9c16f3118206b80b8a0d8", "completed_utc": "2026-08-02T00:24:06.526939+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 890, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.24700889895058, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 890, "prompt_tokens_per_second": 289.9768936133507, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2397843456, "peak_system_memory_used_bytes": 55890763776, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 149.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 144.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 148.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 149.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 145.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 145.96, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 24, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5994291389361024}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "fr", "max_score": 10.0, "passed": false, "score": 5.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 162, "path": "results/quants/UD-IQ3_XXS/cases/LANG-fr/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1440b96075ea82e840089fffa9ee7648e8bbac181c7d1648590fb295171777ed", "utf8": "Mara Chen n'a pas approuvé l'expédition VX-44. Elle reste en attente jusqu'au 17 août 2026. La limite approuvée est de 12 450,75 €, et non de 12 540,75 €."}]}
{"case_id": "LANG-fr", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-fr/seed-2731/checkpoint.json", "checkpoint_sha256": "cfe1fefce6f2facc66faf2f08626448ffd843f040a2673757b5de6ba2b0d7ca0", "completed_utc": "2026-08-02T00:24:13.658903+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 215, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.79846750786162, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 215, "prompt_tokens_per_second": 290.22297574066556, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2440941568, "peak_system_memory_used_bytes": 55901474816, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 147.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 144.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 141.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 146.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 141.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 144.01, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5979892769828439}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "fr", "max_score": 10.0, "passed": false, "score": 5.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 156, "path": "results/quants/UD-IQ3_XXS/cases/LANG-fr/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "52d6877e3047f3ddb16c831e6fc99e67c5cd2c482cdfa69e9d985d1aee96c5e0", "utf8": "Mara Chen n'a pas approuvé l'expédition VX-44. Elle reste en attente jusqu'au 17 août 2026. La limite approuvée est de 12 450,75 €, pas 12 540,75 €."}]}
{"case_id": "LANG-fr", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-fr/seed-731/checkpoint.json", "checkpoint_sha256": "b4140402231420f71cbe147ca33f90ca49215b8f23bc5e687568675d4b5e0f06", "completed_utc": "2026-08-02T00:23:39.647546+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 969, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.235084517379114, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 969, "prompt_tokens_per_second": 289.7232592766454, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2354032640, "peak_system_memory_used_bytes": 55813853184, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 147.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 143.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 144.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 145.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 156.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 157.95, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 26, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5968356369994581}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 0.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "fr", "max_score": 10.0, "passed": false, "score": 5.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 156, "path": "results/quants/UD-IQ3_XXS/cases/LANG-fr/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "52d6877e3047f3ddb16c831e6fc99e67c5cd2c482cdfa69e9d985d1aee96c5e0", "utf8": "Mara Chen n'a pas approuvé l'expédition VX-44. Elle reste en attente jusqu'au 17 août 2026. La limite approuvée est de 12 450,75 €, pas 12 540,75 €."}]}
{"case_id": "LANG-ja", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-ja/seed-1731/checkpoint.json", "checkpoint_sha256": "2b363c6e83607b23248c109f9d864659e15d227ec9591b45efd1a9dc4d0dad06", "completed_utc": "2026-08-02T00:24:45.478189+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 221, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.83735055801239, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 221, "prompt_tokens_per_second": 290.02799320820526, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2522202112, "peak_system_memory_used_bytes": 55906619392, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 147.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 144.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 139.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 144.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 142.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 143.52, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5968316519865766}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 0.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "ja", "max_score": 10.0, "passed": false, "score": 6.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 201, "path": "results/quants/UD-IQ3_XXS/cases/LANG-ja/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9b17c244e10d9702cf9a2ce7216d59ea3155d3ae8b2d4f90eb0165d5da00502d", "utf8": "Mara Chenはshipment VX-44を承認しませんでした。それは2026年8月17日まで保留のままです。承認された限度額は€12,450.75であり、€12,540.75ではありません。"}]}
{"case_id": "LANG-ja", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-ja/seed-2731/checkpoint.json", "checkpoint_sha256": "27a3de4a594d71c079d147d88c655da89eab4e6e1192e0888a6b03d21d3aa277", "completed_utc": "2026-08-02T00:24:53.658795+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 250, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.68569144469096, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 250, "prompt_tokens_per_second": 291.92225260512896, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2560708608, "peak_system_memory_used_bytes": 55925051392, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 147.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 149.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 147.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 162.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 160.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 144.55, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5755949640879408}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 0.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "ja", "max_score": 10.0, "passed": false, "score": 6.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 195, "path": "results/quants/UD-IQ3_XXS/cases/LANG-ja/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "45698b33f4334820a6d2de6579012e61a1d7e6fafb6f412b348b85f19b7bf32b", "utf8": "Mara Chenは出荷VX-44を承認しませんでした。それは2026年8月17日まで保留のままです。承認済みの上限は€12,450.75であり、€12,540.75ではありません。"}]}
{"case_id": "LANG-ja", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-ja/seed-731/checkpoint.json", "checkpoint_sha256": "e58bc3c30e140123ec1d1ddac9782a1bf89781cc89dd89165f01717eae673cf9", "completed_utc": "2026-08-02T00:24:38.250686+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 810, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.356093014922756, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 810, "prompt_tokens_per_second": 289.9077249399543, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2479611904, "peak_system_memory_used_bytes": 55979577344, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 149.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 145.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 139.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 150.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 141.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 143.87, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 22, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5948830579873174}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 1.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "ja", "max_score": 10.0, "passed": false, "score": 7.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 204, "path": "results/quants/UD-IQ3_XXS/cases/LANG-ja/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "840aafa078992153299a3fb87dbb51a4dc4eac30229e02d1867eedfc7934aaa4", "utf8": "Mara Chen は出荷 VX-44 を承認しませんでした。それは2026年8月17日まで保留のままです。承認された限度額は €12,450.75 であり、€12,540.75 ではありません。"}]}
{"case_id": "LANG-zh-Hans", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-zh-Hans/seed-1731/checkpoint.json", "checkpoint_sha256": "68d8c49348e33cabc8ecb26b7696b58a5a97f4ccce733e8f716925a8bb54f1ab", "completed_utc": "2026-08-02T00:25:15.839806+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 507, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.52305782775609, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 507, "prompt_tokens_per_second": 292.6777835747774, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2637803520, "peak_system_memory_used_bytes": 56033828864, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 147.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 142.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 149.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 149.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 144.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 145.35, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5972864889772609}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 0.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "zh-Hans", "max_score": 10.0, "passed": false, "score": 6.0, "scorer": "multilingual_fact_preservation"}, "seed": 1731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 142, "path": "results/quants/UD-IQ3_XXS/cases/LANG-zh-Hans/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "651facba0003a73fbbd684bdd48dae3d6d4e39593730a1339bcdf716a9ff8226", "utf8": "Mara Chen未批准发货VX-44。该货件仍保持搁置状态，直到17 August 2026。批准的限额是€12,450.75，不是€12,540.75。"}]}
{"case_id": "LANG-zh-Hans", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-zh-Hans/seed-2731/checkpoint.json", "checkpoint_sha256": "3df6e86e74190620dc9711b423419a08f9254dc5b7bf2dba9bd746547a12b174", "completed_utc": "2026-08-02T00:25:39.908200+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 793, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.36192037116767, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 793, "prompt_tokens_per_second": 292.7996741012323, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2678431744, "peak_system_memory_used_bytes": 56101163008, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 147.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 143.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 139.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 144.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 143.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 144.58, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 21, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5994838560000062}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 0.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "zh-Hans", "max_score": 10.0, "passed": false, "score": 6.0, "scorer": "multilingual_fact_preservation"}, "seed": 2731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 140, "path": "results/quants/UD-IQ3_XXS/cases/LANG-zh-Hans/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "f9be2b5adb88d336ff1d9aeb9df15b3769579d1d28d7bb0e9f1237c51e551356", "utf8": "Mara Chen 未批准货运 VX-44。该货运保持搁置状态直至 2026年8月17日。批准限额为 €12,450.75，而非 €12,540.75。"}]}
{"case_id": "LANG-zh-Hans", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/LANG-zh-Hans/seed-731/checkpoint.json", "checkpoint_sha256": "231d6f0ea92cc3d8abd0987346301e78e9690ab70a918a68bc314f18b7b6bd77", "completed_utc": "2026-08-02T00:25:00.113408+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 1024, "output_tokens": 191, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": false, "diagnostic_only": true, "performance": {"decode_tokens_per_second": 34.830192780558114, "finish_reason": "stop", "output_token_cap": 1024, "output_tokens": 191, "prompt_tokens_per_second": 291.90780789927913, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2599436288, "peak_system_memory_used_bytes": 56039550976, "per_gpu": {"0": {"peak_memory_used_mib": 19791.0, "peak_power_draw_w": 145.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18225.0, "peak_power_draw_w": 142.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17811.0, "peak_power_draw_w": 139.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 21989.0, "peak_power_draw_w": 147.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19377.0, "peak_power_draw_w": 141.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16805.0, "peak_power_draw_w": 142.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.5979403540259227}, "quant": "UD-IQ3_XXS", "score": {"canonical_exact": false, "components": {"both_amounts": 2.0, "date": 1.0, "hold_status": 0.0, "name": 1.0, "negation": 1.0, "semantic_adequacy": 0.0, "shipment_identifier": 1.0}, "language": "zh-Hans", "max_score": 10.0, "passed": false, "score": 6.0, "scorer": "multilingual_fact_preservation"}, "seed": 731, "status": "completed", "task": "multilingual", "visible_model_outputs": [{"bytes": 151, "path": "results/quants/UD-IQ3_XXS/cases/LANG-zh-Hans/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "621ef53496ea5edda4c2bc17f6faebc2a204792dd3cce72ce39ac8e9bc965736", "utf8": "Mara Chen 未批准发货 VX-44。该发货仍处于搁置状态，直至 2026 年 8 月 17 日。批准限额为 €12,450.75，而非 €12,540.75。"}]}
{"case_id": "PY-PATCH-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/PY-PATCH-01/seed-1731/checkpoint.json", "checkpoint_sha256": "f9cab83b8cac83f35fe4303f6dd2eebeb4e30c25557c0908bec88f34fea8b095", "completed_utc": "2026-08-02T00:29:58.057206+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.226071683481884, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 600.1166062872002, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2809565184, "peak_system_memory_used_bytes": 56254435328, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 159.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 150.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 151.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22015.0, "peak_power_draw_w": 157.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 150.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 165.72, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 117, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4919638340361416}, "quant": "UD-IQ3_XXS", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "python_patch"}, "seed": 1731, "status": "completed", "task": "python_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/PY-PATCH-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "PY-PATCH-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/PY-PATCH-01/seed-2731/checkpoint.json", "checkpoint_sha256": "a81bc1be0437d934f30f963aae78f0664361c51dfdbd08e1fbb5cf6ffab5c6b1", "completed_utc": "2026-08-02T00:32:07.044325+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.223708565255265, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 601.588965588112, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2879676416, "peak_system_memory_used_bytes": 57368997888, "per_gpu": {"0": {"peak_memory_used_mib": 20086.0, "peak_power_draw_w": 163.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18520.0, "peak_power_draw_w": 144.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 18106.0, "peak_power_draw_w": 153.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22282.0, "peak_power_draw_w": 171.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19420.0, "peak_power_draw_w": 156.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 150.99, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 116, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4924301919527352}, "quant": "UD-IQ3_XXS", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "python_patch"}, "seed": 2731, "status": "completed", "task": "python_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/PY-PATCH-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "PY-PATCH-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/PY-PATCH-01/seed-731/checkpoint.json", "checkpoint_sha256": "614280a25fa5e74ec4141ffdfd3cd3d4500a076f740b0a21fe547146b3225447", "completed_utc": "2026-08-02T00:27:49.084332+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.18053203045966, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 601.5327310870402, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2739445760, "peak_system_memory_used_bytes": 56191909888, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 172.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 151.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 146.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22015.0, "peak_power_draw_w": 149.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 162.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 167.22, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 115, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4590796800330281}, "quant": "UD-IQ3_XXS", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "python_patch"}, "seed": 731, "status": "completed", "task": "python_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/PY-PATCH-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-CF-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-01/seed-1731/checkpoint.json", "checkpoint_sha256": "c676b9cf54cc9bd6f633168882b422a6930be84069c640cecdfc3299beaa93f7", "completed_utc": "2026-08-02T00:32:41.228563+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 476, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.44283896113162, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 476, "prompt_tokens_per_second": 304.05640913830416, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2953637888, "peak_system_memory_used_bytes": 56423104512, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 142.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 149.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 149.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 144.04, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7203715189825743}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 31, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a6100b4182517cae95ba792e4147ca4d36ff0eb0a03438e07ad7de1c276bde52", "utf8": "[\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P9\"]"}]}
{"case_id": "R-CF-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-01/seed-2731/checkpoint.json", "checkpoint_sha256": "f32c6fb817db5ee66ecc0ea4dfbf2bfd39228a1b0aa7e1c8afb019dc3a9cfe0f", "completed_utc": "2026-08-02T00:32:57.683366+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 527, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.41415548903168, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 527, "prompt_tokens_per_second": 304.4895121929483, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2994290688, "peak_system_memory_used_bytes": 56382222336, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 140.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 142.67, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7261575440643355}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "76061ca5fbef0891168215924b8e409ae5f5476fc3e95cfa5873ba807226fc13", "utf8": "[\"P3\", \"P4\", \"P5\", \"P6\", \"P7\", \"P9\"]"}]}
{"case_id": "R-CF-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-01/seed-731/checkpoint.json", "checkpoint_sha256": "f5e661957d40b5c457b25874b556f67b0fb5145fdd8b056f823f9bec5ba2edc9", "completed_utc": "2026-08-02T00:32:26.306830+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 621, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.37219082754011, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 621, "prompt_tokens_per_second": 305.31986351457414, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 2929844224, "peak_system_memory_used_bytes": 56344547328, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 143.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 146.05, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 17, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.764045394025743}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 31, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a6100b4182517cae95ba792e4147ca4d36ff0eb0a03438e07ad7de1c276bde52", "utf8": "[\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P9\"]"}]}
{"case_id": "R-CF-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-02/seed-1731/checkpoint.json", "checkpoint_sha256": "0db917afafe9c8ef95823d6e3c5fb06db73dca56dfeb54569f50b03b945a8b37", "completed_utc": "2026-08-02T00:33:33.107237+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 456, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.422267160726854, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 456, "prompt_tokens_per_second": 305.1653326227695, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3077349376, "peak_system_memory_used_bytes": 56502255616, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 159.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 142.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 148.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 140.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 143.19, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 12, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7262119050137699}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-02/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "155ba39cf2c1ba52f3d26c7f08cad012d6a2fe8e0df297ae8c0ceaff5469ce73", "utf8": "[\"P1\",\"P3\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-02/seed-2731/checkpoint.json", "checkpoint_sha256": "e06ebcf79550bb57918b4ce4e190984b9621a75780ecf11bec379a3b5f66b5ae", "completed_utc": "2026-08-02T00:33:55.941361+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 742, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.17366781647396, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 742, "prompt_tokens_per_second": 300.54596740615636, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3117875200, "peak_system_memory_used_bytes": 56712646656, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 141.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 143.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 150.9, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 17, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7408531879773363}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-02/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "155ba39cf2c1ba52f3d26c7f08cad012d6a2fe8e0df297ae8c0ceaff5469ce73", "utf8": "[\"P1\",\"P3\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-02/seed-731/checkpoint.json", "checkpoint_sha256": "f5b4fc4a0f5c12b0e6c7624d919441747e2dad58439a9c7a55be09822cb49f51", "completed_utc": "2026-08-02T00:33:18.759219+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 685, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.333500790597775, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 685, "prompt_tokens_per_second": 306.1639007371829, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3035275264, "peak_system_memory_used_bytes": 56453787648, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 149.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 144.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 144.46, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 18, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7231278929393739}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-02/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "155ba39cf2c1ba52f3d26c7f08cad012d6a2fe8e0df297ae8c0ceaff5469ce73", "utf8": "[\"P1\",\"P3\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-03/seed-1731/checkpoint.json", "checkpoint_sha256": "fbe5e89000e6fae553d4368296a4a3c9641e168829ecde5e4bd1e2c5e9b99fb0", "completed_utc": "2026-08-02T00:34:29.785401+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 426, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.53621791283366, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 426, "prompt_tokens_per_second": 302.44591907625227, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3202265088, "peak_system_memory_used_bytes": 56591306752, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 143.14, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 11, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7250036160694435}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 26, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-03/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "19615736188f8934441742266c5022e401ff7a5834af876662e52d8d56d5099f", "utf8": "[\"P3\",\"P4\",\"P5\",\"P6\",\"P9\"]"}]}
{"case_id": "R-CF-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-03/seed-2731/checkpoint.json", "checkpoint_sha256": "47eb51b7ca9a9bee321267c7305ad024df5768227d130885d5d5255721ad0c99", "completed_utc": "2026-08-02T00:34:40.859007+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 344, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.502330461917, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 344, "prompt_tokens_per_second": 302.05788146770226, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3242516480, "peak_system_memory_used_bytes": 56624590848, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 145.02, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 10, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7157481310423464}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 26, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-03/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "19615736188f8934441742266c5022e401ff7a5834af876662e52d8d56d5099f", "utf8": "[\"P3\",\"P4\",\"P5\",\"P6\",\"P9\"]"}]}
{"case_id": "R-CF-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-03/seed-731/checkpoint.json", "checkpoint_sha256": "83b75e2ca242d765e03937a1841f1882f4861f661ecd231768e860d5540df999", "completed_utc": "2026-08-02T00:34:16.357220+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 663, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.4127839911106, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 663, "prompt_tokens_per_second": 302.5936167726743, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3160358912, "peak_system_memory_used_bytes": 56582811648, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 148.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 144.17, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 18, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.719919175026007}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 26, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-03/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "19615736188f8934441742266c5022e401ff7a5834af876662e52d8d56d5099f", "utf8": "[\"P3\",\"P4\",\"P5\",\"P6\",\"P9\"]"}]}
{"case_id": "R-CF-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-04/seed-1731/checkpoint.json", "checkpoint_sha256": "84314ab3751505924b28c7b51d63ce0db62d3793be8ea4eea6cc699de79a7e87", "completed_utc": "2026-08-02T00:35:06.046600+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 322, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.48279924082156, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 322, "prompt_tokens_per_second": 305.3839943034713, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3322806272, "peak_system_memory_used_bytes": 56648433664, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 142.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 146.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 151.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 154.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7218124610371888}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-04/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-04/seed-2731/checkpoint.json", "checkpoint_sha256": "bf74a062310ce4f10606e59e8e7e77b4cdabef9ca1917aaecea43bc41175094d", "completed_utc": "2026-08-02T00:35:22.276540+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 520, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.47422693534995, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 520, "prompt_tokens_per_second": 304.9283864528499, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3362328576, "peak_system_memory_used_bytes": 56720732160, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 150.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 146.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 144.78, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7299299669684842}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-04/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-04/seed-731/checkpoint.json", "checkpoint_sha256": "4440027c33f1ef7e953e169ad3cd79f364b1f2bfbea206a0fae5972c6f35d022", "completed_utc": "2026-08-02T00:34:55.567464+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 469, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.45126336235222, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 469, "prompt_tokens_per_second": 305.2921273351124, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3282231296, "peak_system_memory_used_bytes": 56620122112, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 143.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 152.33, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.72182466997765}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-04/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-05/seed-1731/checkpoint.json", "checkpoint_sha256": "1d407658055fc5c8d649be12b428e7f501ebcf5bd54b2571395f40e4a1e94b37", "completed_utc": "2026-08-02T00:36:01.233427+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 691, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.40363489588071, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 691, "prompt_tokens_per_second": 304.70947066762585, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3444355072, "peak_system_memory_used_bytes": 56868237312, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 165.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 145.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 150.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 153.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 147.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 146.48, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 19, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.724304897012189}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-05/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b7c5d63364adfb6a5a28ccff4ba1ab0a92bb532e4161877d32f5177838853194", "utf8": "[\"P1\",\"P3\",\"P4\",\"P5\",\"P6\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-05/seed-2731/checkpoint.json", "checkpoint_sha256": "358813e7ebd45ba3bda93fc750d82b5a0990c3f6c67989bff8859aa2d39b0dfa", "completed_utc": "2026-08-02T00:36:15.224165+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 444, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.443246257341784, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 444, "prompt_tokens_per_second": 304.68818508395276, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3486339072, "peak_system_memory_used_bytes": 56918786048, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 142.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 145.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 143.29, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7308432159479707}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-05/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b7c5d63364adfb6a5a28ccff4ba1ab0a92bb532e4161877d32f5177838853194", "utf8": "[\"P1\",\"P3\",\"P4\",\"P5\",\"P6\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-05/seed-731/checkpoint.json", "checkpoint_sha256": "80e3b07560f05e7a4645d06ecd660e6b92cbcfbf6ff9410421a7370563180a3d", "completed_utc": "2026-08-02T00:35:40.020435+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 571, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.417419771246514, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 571, "prompt_tokens_per_second": 304.63339970874074, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3403165696, "peak_system_memory_used_bytes": 56801812480, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 142.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 151.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 145.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 158.8, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 16, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7245405869325623}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 36, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-05/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b7c5d63364adfb6a5a28ccff4ba1ab0a92bb532e4161877d32f5177838853194", "utf8": "[\"P1\",\"P3\",\"P4\",\"P5\",\"P6\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-06/seed-1731/checkpoint.json", "checkpoint_sha256": "c09b0f34a75bb9f0969b3128df1c32ac2125ed9494446fadc436bb4db0b792cd", "completed_utc": "2026-08-02T00:36:52.098936+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 563, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.42847057176027, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 563, "prompt_tokens_per_second": 307.8895209505497, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3568279552, "peak_system_memory_used_bytes": 57004392448, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 149.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 145.05, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 15, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7285123090259731}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-06/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-06/seed-2731/checkpoint.json", "checkpoint_sha256": "330cc4e2418964bd7305f2b521152df53cc69f1aaccc3d60b82a070a41728eca", "completed_utc": "2026-08-02T00:37:09.356764+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 555, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.49399301095766, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 555, "prompt_tokens_per_second": 308.0810549406277, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3609436160, "peak_system_memory_used_bytes": 57082249216, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 149.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 144.0, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 15, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7320334900869057}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-06/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-06/seed-731/checkpoint.json", "checkpoint_sha256": "8e0dc236a79a9b627e20f9e1263a58586c4bea01d4249c35e84efdf5640aee35", "completed_utc": "2026-08-02T00:36:34.633869+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 630, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.39863353339324, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 630, "prompt_tokens_per_second": 307.892711232163, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3526672384, "peak_system_memory_used_bytes": 56864935936, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 152.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 149.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 144.17, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 17, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7284914130577818}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 46, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-06/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9340ab5a15656889a12ee18eef7d80a2ba775f12dd971e12fb33da3949dd43ed", "utf8": "[\"P1\",\"P2\",\"P3\",\"P4\",\"P5\",\"P6\",\"P7\",\"P8\",\"P9\"]"}]}
{"case_id": "R-CF-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-07/seed-1731/checkpoint.json", "checkpoint_sha256": "fe130590568ff87b77f835000faadbe65c40ba2f285fd7e8ea5d0774872aa9f3", "completed_utc": "2026-08-02T00:37:44.198268+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 555, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.47068156735891, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 555, "prompt_tokens_per_second": 314.8925024683747, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3691724800, "peak_system_memory_used_bytes": 57068257280, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 143.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 154.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 155.73, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 15, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7323479480110109}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 30, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-07/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "909bfa8687c5ddbc479433905568f34e553d34b1de20fb6cab8dd83bfc413fd7", "utf8": "[\"P1\", \"P2\", \"P4\", \"P5\", \"P8\"]"}]}
{"case_id": "R-CF-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-07/seed-2731/checkpoint.json", "checkpoint_sha256": "af25bc7ba3368e1a409237844e120fc54cb4ce237cdb969f4fe825e0cfee2bd8", "completed_utc": "2026-08-02T00:38:00.429383+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 519, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.478621461008665, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 519, "prompt_tokens_per_second": 314.2917567440549, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3732873216, "peak_system_memory_used_bytes": 57117024256, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 144.97, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7380636420566589}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 26, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-07/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "6438a9b502d71fb4204373a8da88f5cd7a0e7adb664e123ec8af51a8f769399c", "utf8": "[\"P1\",\"P2\",\"P4\",\"P5\",\"P8\"]"}]}
{"case_id": "R-CF-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-CF-07/seed-731/checkpoint.json", "checkpoint_sha256": "079736dbf6ba0a8654f86433c539d315c8edbb9b96cd425a24693818db5aa64e", "completed_utc": "2026-08-02T00:37:26.932570+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 565, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.41425147145289, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 565, "prompt_tokens_per_second": 314.3882577454836, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3650519040, "peak_system_memory_used_bytes": 57018175488, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 164.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 148.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 148.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 155.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 147.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 145.82, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 16, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7433030550600961}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 30, "path": "results/quants/UD-IQ3_XXS/cases/R-CF-07/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "909bfa8687c5ddbc479433905568f34e553d34b1de20fb6cab8dd83bfc413fd7", "utf8": "[\"P1\", \"P2\", \"P4\", \"P5\", \"P8\"]"}]}
{"case_id": "R-GRAPH-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-01/seed-1731/checkpoint.json", "checkpoint_sha256": "157fc4238fcd672a7dfb9c91bff2aeec8fe96c40b99b426ccc08607620438a6a", "completed_utc": "2026-08-02T00:38:14.242585+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 255, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.611535060942096, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 255, "prompt_tokens_per_second": 297.9656288730134, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3773857792, "peak_system_memory_used_bytes": 57097814016, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 149.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 146.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 153.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 152.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 151.58, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7104699009796605}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "80bed85c4ee03d4fc0818522004a326c9facf394a916a487a3744b49804b99a2", "utf8": "{\"reachable\":[\"E\",\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-01/seed-2731/checkpoint.json", "checkpoint_sha256": "318f1e42e66171cec7cc1742598ebdb64184298f82f237f442450b1bb3ea891a", "completed_utc": "2026-08-02T00:38:20.043245+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 162, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.72485695288094, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 162, "prompt_tokens_per_second": 301.08565377767957, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3812851712, "peak_system_memory_used_bytes": 57111949312, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 162.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 140.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 143.55, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7420410490594804}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "80bed85c4ee03d4fc0818522004a326c9facf394a916a487a3744b49804b99a2", "utf8": "{\"reachable\":[\"E\",\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-01/seed-731/checkpoint.json", "checkpoint_sha256": "7fc36843c14efe447d07231b03d7ea5dcd34c1b4f6e5f97552f872c8013612cd", "completed_utc": "2026-08-02T00:38:05.761631+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 145, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.76435402196004, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 145, "prompt_tokens_per_second": 302.05044096446017, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3773734912, "peak_system_memory_used_bytes": 57108725760, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 151.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 148.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 150.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 150.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 145.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16833.0, "peak_power_draw_w": 146.12, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7414028000785038}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "80bed85c4ee03d4fc0818522004a326c9facf394a916a487a3744b49804b99a2", "utf8": "{\"reachable\":[\"E\",\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-02/seed-1731/checkpoint.json", "checkpoint_sha256": "c0a3cb869df2e7baa5841e7961d35f5c4400a1b5abdfcc7af52399bf161492c1", "completed_utc": "2026-08-02T00:38:28.898977+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 117, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.74880770009819, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 117, "prompt_tokens_per_second": 305.1381429068961, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3851313152, "peak_system_memory_used_bytes": 57209540608, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 143.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 153.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 150.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 148.84, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6423569680191576}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 44, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-02/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "027d14d5a4eca73d6cd0cc389f7448a0a3c7d3ac648fe299e67b153bf55832af", "utf8": "{\"reachable\":[\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-02/seed-2731/checkpoint.json", "checkpoint_sha256": "e32e484dbd2780ab70f18f889d0b9fa4b72cc5d0bf4cfb4a454365cc22ab4b63", "completed_utc": "2026-08-02T00:38:33.700298+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 130, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.765670024049825, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 130, "prompt_tokens_per_second": 305.2805544278467, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3851436032, "peak_system_memory_used_bytes": 57268334592, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 156.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 145.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 155.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 151.97, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6397165149683133}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 44, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-02/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "027d14d5a4eca73d6cd0cc389f7448a0a3c7d3ac648fe299e67b153bf55832af", "utf8": "{\"reachable\":[\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-02/seed-731/checkpoint.json", "checkpoint_sha256": "029243e4c77b8d9124764a52a4d32752cbe6f9826c15246cb83b3cf011780bd7", "completed_utc": "2026-08-02T00:38:24.535745+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 119, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.80424658603975, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 119, "prompt_tokens_per_second": 305.71103424763874, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3851190272, "peak_system_memory_used_bytes": 57168420864, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 151.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 153.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 147.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 151.23, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6742222069296986}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 44, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-02/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "027d14d5a4eca73d6cd0cc389f7448a0a3c7d3ac648fe299e67b153bf55832af", "utf8": "{\"reachable\":[\"F\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-03/seed-1731/checkpoint.json", "checkpoint_sha256": "b573a63384518376154d15c297dd160c7d35d33a5b688eec82918e03b0129b93", "completed_utc": "2026-08-02T00:38:49.483700+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 210, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.50516714878053, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 210, "prompt_tokens_per_second": 295.93648590730044, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3928436736, "peak_system_memory_used_bytes": 57441615872, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 146.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 151.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 149.18, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7357856440357864}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-03/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b22e387e29cad42a2cecb3c833850242fe106aa67d7430530bac5d1e558d4d68", "utf8": "{\"reachable\":[\"E\",\"F\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-03/seed-2731/checkpoint.json", "checkpoint_sha256": "6cfa52a8e204da223947424840244864213fb9f94abc9dba7f48b6230d18c580", "completed_utc": "2026-08-02T00:38:57.196536+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 228, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.58406293571369, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 228, "prompt_tokens_per_second": 295.7239482497658, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3967119360, "peak_system_memory_used_bytes": 57426440192, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 146.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 145.75, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.737274543964304}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-03/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b22e387e29cad42a2cecb3c833850242fe106aa67d7430530bac5d1e558d4d68", "utf8": "{\"reachable\":[\"E\",\"F\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-03/seed-731/checkpoint.json", "checkpoint_sha256": "f4643c6075aeea41d57614fa95a66d3d211c1fddb57fd7ac16983ab3a5196bcd", "completed_utc": "2026-08-02T00:38:42.321614+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 258, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.54552417209241, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 258, "prompt_tokens_per_second": 295.9623734141228, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 3889455104, "peak_system_memory_used_bytes": 57431920640, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 146.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 143.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.8, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7340659330366179}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-03/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b22e387e29cad42a2cecb3c833850242fe106aa67d7430530bac5d1e558d4d68", "utf8": "{\"reachable\":[\"E\",\"F\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-04/seed-1731/checkpoint.json", "checkpoint_sha256": "2acf94f0188555f83a7d65f6ea60333f792352a0a37f9b120e26ec292f7e6c57", "completed_utc": "2026-08-02T00:39:11.418878+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 168, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.728308863656665, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 168, "prompt_tokens_per_second": 292.6969894864418, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4044845056, "peak_system_memory_used_bytes": 57420361728, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 146.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 141.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 146.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 154.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 147.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 152.23, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7349694430595264}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 47, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-04/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8ae9db64671b10d5aca89092eda3e058a63c807d44ddf14e23414eee8a384764", "utf8": "{\"reachable\":[\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-04/seed-2731/checkpoint.json", "checkpoint_sha256": "75579beb267fa7d03dc214fffd68377c16bc7777497d6cdfb735c1c1de777488", "completed_utc": "2026-08-02T00:39:18.013218+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 191, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.54223937156996, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 191, "prompt_tokens_per_second": 291.5823544322716, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4044972032, "peak_system_memory_used_bytes": 57391165440, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.23, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7001052720006555}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 47, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-04/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8ae9db64671b10d5aca89092eda3e058a63c807d44ddf14e23414eee8a384764", "utf8": "{\"reachable\":[\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-04/seed-731/checkpoint.json", "checkpoint_sha256": "84b2f5cefd49b610e119eef6170c7c651394e8022e134397f2657ddfa741085c", "completed_utc": "2026-08-02T00:39:05.424165+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 245, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.55980192860622, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 245, "prompt_tokens_per_second": 293.2253163960103, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4005974016, "peak_system_memory_used_bytes": 57423720448, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 149.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 143.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 143.75, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7286163059761748}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 47, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-04/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8ae9db64671b10d5aca89092eda3e058a63c807d44ddf14e23414eee8a384764", "utf8": "{\"reachable\":[\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-05/seed-1731/checkpoint.json", "checkpoint_sha256": "a8709df9a6192928c56f75bd3f8414c64ad3ba1fd0876073b05b2ee964881f14", "completed_utc": "2026-08-02T00:39:38.612849+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 270, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.54033024905241, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 270, "prompt_tokens_per_second": 322.17866581130835, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4123385856, "peak_system_memory_used_bytes": 57494573056, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 146.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 143.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 155.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 156.17, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7454371029743925}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-05/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fe0eb6dafe2fb00aee4831bc12e5fea8af950087540db3bed2ad0b4d1bd8eab1", "utf8": "{\"reachable\":[\"E\",\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-05/seed-2731/checkpoint.json", "checkpoint_sha256": "7eca7f6322a00a29a07fee07211889ec79ccc1b54571dfc7540b67adf3b4c346", "completed_utc": "2026-08-02T00:39:49.059180+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 322, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.43284243007326, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 322, "prompt_tokens_per_second": 316.87137674331603, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4162637824, "peak_system_memory_used_bytes": 57524776960, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 150.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 143.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 155.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.15, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7656428680056706}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-05/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fe0eb6dafe2fb00aee4831bc12e5fea8af950087540db3bed2ad0b4d1bd8eab1", "utf8": "{\"reachable\":[\"E\",\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-05/seed-731/checkpoint.json", "checkpoint_sha256": "5b9fbfd5a2a3f5a48a8f506b673b188571832e40af62b1d6bba166b0182e953f", "completed_utc": "2026-08-02T00:39:29.668510+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 363, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.46740594030074, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 363, "prompt_tokens_per_second": 320.5775479913063, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4083507200, "peak_system_memory_used_bytes": 57521184768, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 164.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 148.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 148.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.46, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 11, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7622414430370554}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 51, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-05/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fe0eb6dafe2fb00aee4831bc12e5fea8af950087540db3bed2ad0b4d1bd8eab1", "utf8": "{\"reachable\":[\"E\",\"G\",\"H\"],\"target_reachable\":true}"}]}
{"case_id": "R-GRAPH-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-06/seed-1731/checkpoint.json", "checkpoint_sha256": "78332383ae1360b246927ae36ff1e18ba8e4252aa53dea12c2735105eda31dc2", "completed_utc": "2026-08-02T00:40:04.034734+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 228, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.489784897195484, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 228, "prompt_tokens_per_second": 306.81492661888495, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4241010688, "peak_system_memory_used_bytes": 57609338880, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 148.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 151.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 145.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 153.8, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7266230980167165}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 56, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-06/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ab97d253da1946994b0f7c0ae1d8bc073294787350658f34ac8d55d3fe98ee24", "utf8": "{\"reachable\":[\"D\",\"E\",\"F\",\"G\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-06/seed-2731/checkpoint.json", "checkpoint_sha256": "2fe2fefad982c569a423e55abedfe67246ba974e03596a2d208f7e6c13acb85c", "completed_utc": "2026-08-02T00:40:10.516360+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 189, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.47534903098732, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 189, "prompt_tokens_per_second": 306.9265020124778, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4279836672, "peak_system_memory_used_bytes": 57655758848, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 153.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 144.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 140.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 142.24, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7296149629401043}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 56, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-06/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ab97d253da1946994b0f7c0ae1d8bc073294787350658f34ac8d55d3fe98ee24", "utf8": "{\"reachable\":[\"D\",\"E\",\"F\",\"G\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-06/seed-731/checkpoint.json", "checkpoint_sha256": "55a1ae5b31f88367a78ec5b1b5ed0491ab409d55064b955475e1e51a0915ad6f", "completed_utc": "2026-08-02T00:39:56.427442+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 218, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.37414458837435, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 218, "prompt_tokens_per_second": 300.5595250304833, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4202237952, "peak_system_memory_used_bytes": 57573781504, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 145.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 141.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 142.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 144.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 149.39, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7543140890775248}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 57, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-06/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e6991b0921e451bca9abeeb462e895bcf964a01bcf46d61510b24f2f8e60c1f2", "utf8": " {\"reachable\":[\"D\",\"E\",\"F\",\"G\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-07/seed-1731/checkpoint.json", "checkpoint_sha256": "9803e322b9164553ee76c5b35c933af48fdacb1bcd1676f7c5db90223e940a2d", "completed_utc": "2026-08-02T00:40:21.120934+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 155, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.68470927947998, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 155, "prompt_tokens_per_second": 302.3062903075054, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4318527488, "peak_system_memory_used_bytes": 57779412992, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.73, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 151.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 145.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 142.71, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.6999259289586917}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 41, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-07/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a7add69bd5722467ba5ca5f5212bc9d53463eeb8315d3b18cc337421f5c0dfa0", "utf8": "{\"reachable\":[],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-07/seed-2731/checkpoint.json", "checkpoint_sha256": "171cbd279e7eeea82dda934cefcd5691e60abc10c75c894ef5f8be5dbd38e91c", "completed_utc": "2026-08-02T00:40:26.646682+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 158, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.704413940192, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 158, "prompt_tokens_per_second": 301.1277014209736, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4318654464, "peak_system_memory_used_bytes": 57760854016, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 149.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 148.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 145.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 148.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 146.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 145.67, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7076387410052121}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 41, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-07/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a7add69bd5722467ba5ca5f5212bc9d53463eeb8315d3b18cc337421f5c0dfa0", "utf8": "{\"reachable\":[],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-07/seed-731/checkpoint.json", "checkpoint_sha256": "b6b7a56bb5187a90ebcca7001b9cc762cee10a17844d61c5f497267300c3ab60", "completed_utc": "2026-08-02T00:40:15.712209+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 146, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.48161031954298, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 146, "prompt_tokens_per_second": 301.8451927872122, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4318396416, "peak_system_memory_used_bytes": 57644564480, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 144.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 147.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 149.36, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 137.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 142.37, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7248462139395997}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 41, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-07/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "a7add69bd5722467ba5ca5f5212bc9d53463eeb8315d3b18cc337421f5c0dfa0", "utf8": "{\"reachable\":[],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-08", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-08/seed-1731/checkpoint.json", "checkpoint_sha256": "1d8ae0b8c46493ba701139de54e6a8f28bcf892957e4bb3bea6f58ea963dc556", "completed_utc": "2026-08-02T00:40:41.938139+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 227, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.43355885544671, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 227, "prompt_tokens_per_second": 305.335116143111, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4395823104, "peak_system_memory_used_bytes": 57854009344, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 171.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 153.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 137.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 152.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 145.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.71, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7358824029797688}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-08/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d5ff5eb60c7028268031852d6b3bd3006c4ec539197f97efd2d9a6c29247110a", "utf8": "{\"reachable\":[\"D\",\"H\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-08", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-08/seed-2731/checkpoint.json", "checkpoint_sha256": "fb030c2cb5145a1fc2d4d118947af180d2fd33c95857e4def8fb00ad5ecbe704", "completed_utc": "2026-08-02T00:40:49.356959+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 224, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.54134491044077, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 224, "prompt_tokens_per_second": 306.2058196520283, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4434644992, "peak_system_memory_used_bytes": 57852354560, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 162.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 147.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 143.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 154.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 149.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 151.7, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7295977179892361}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-08/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d5ff5eb60c7028268031852d6b3bd3006c4ec539197f97efd2d9a6c29247110a", "utf8": "{\"reachable\":[\"D\",\"H\"],\"target_reachable\":false}"}]}
{"case_id": "R-GRAPH-08", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-08/seed-731/checkpoint.json", "checkpoint_sha256": "d44333b37226c49dc32331072d4adf4781e5115e5e28c9dbb35e5f5c75c92ba1", "completed_utc": "2026-08-02T00:40:34.346091+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 231, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.39391665415979, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 231, "prompt_tokens_per_second": 306.1067572703982, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4356972544, "peak_system_memory_used_bytes": 57812770816, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 154.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 149.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 152.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 145.32, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7275999210542068}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 48, "path": "results/quants/UD-IQ3_XXS/cases/R-GRAPH-08/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d5ff5eb60c7028268031852d6b3bd3006c4ec539197f97efd2d9a6c29247110a", "utf8": "{\"reachable\":[\"D\",\"H\"],\"target_reachable\":false}"}]}
{"case_id": "R-INV-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-01/seed-1731/checkpoint.json", "checkpoint_sha256": "82b7e840cc14fc670c0f5f612f0677c15a7405bc61116ce20df3b868e4212f50", "completed_utc": "2026-08-02T00:42:54.297886+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.41106517705956, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 308.856159612728, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4525137920, "peak_system_memory_used_bytes": 58173222912, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 164.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 148.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 146.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 150.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 144.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 151.25, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8325541339581832}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-01/seed-2731/checkpoint.json", "checkpoint_sha256": "2701dd5298786a3ddcaf1841132b9fb32ce17f470a342852bc04c08478f48423", "completed_utc": "2026-08-02T00:43:56.746570+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.366217304528384, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 308.2701530000232, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4576763904, "peak_system_memory_used_bytes": 59372433408, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 154.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 145.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 143.17, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8319304320029914}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-01/seed-731/checkpoint.json", "checkpoint_sha256": "169dd5cede49816b789191c4cdea972daf07fb8f3f2aa09f5e64cafb0175adca", "completed_utc": "2026-08-02T00:41:51.886975+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.32846371063265, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 310.4831585620602, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4473511936, "peak_system_memory_used_bytes": 59544379392, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 162.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 143.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 146.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 144.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.68, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8117629470070824}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-02/seed-1731/checkpoint.json", "checkpoint_sha256": "56c6a0c6d77db39c6ce573e0d5b5665a46b99b5cf7bcbe42d902d12d85f2d8b9", "completed_utc": "2026-08-02T00:46:01.453183+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.33098161124211, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 310.23892291276707, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4680015872, "peak_system_memory_used_bytes": 59676274688, "per_gpu": {"0": {"peak_memory_used_mib": 20086.0, "peak_power_draw_w": 149.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18520.0, "peak_power_draw_w": 145.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 18106.0, "peak_power_draw_w": 145.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22284.0, "peak_power_draw_w": 151.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19672.0, "peak_power_draw_w": 149.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 17102.0, "peak_power_draw_w": 148.26, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8256796859204769}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-02/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-02/seed-2731/checkpoint.json", "checkpoint_sha256": "b4ac858a24322e753706ef28be5ff7a942f67043bb2ab15213ae7b05c37a645b", "completed_utc": "2026-08-02T00:47:03.871655+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.446327373599544, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 309.6119616443893, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4731670528, "peak_system_memory_used_bytes": 58375872512, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 166.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 148.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 146.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.01, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 57, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8263086740626022}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-02/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-02/seed-731/checkpoint.json", "checkpoint_sha256": "a0e227cff9810a6cde76a0c827b36416a8ae924df677e02cb5b9c4372b67f327", "completed_utc": "2026-08-02T00:44:59.000401+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.48543207938729, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 310.211943549214, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4628393984, "peak_system_memory_used_bytes": 58164375552, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 171.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 145.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 146.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 146.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 149.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 147.63, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8295936359791085}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-02/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-03/seed-1731/checkpoint.json", "checkpoint_sha256": "205387063d2dcbc8cdd6d07dabed7da52b66f4d58c2bc8822226184e00992ffd", "completed_utc": "2026-08-02T00:49:08.823820+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.44926956872508, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 310.855085602207, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4834967552, "peak_system_memory_used_bytes": 58586615808, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 143.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 147.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 148.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 145.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8249624189920723}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-03/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-03/seed-2731/checkpoint.json", "checkpoint_sha256": "eebdf63e546adb26c5e8f1f34a4b354cb3be974d7671f74799c7b45fa50bd55a", "completed_utc": "2026-08-02T00:50:11.540888+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.2771407864672, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 306.41536333297864, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4886601728, "peak_system_memory_used_bytes": 59912007680, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 158.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 142.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 145.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8582216040231287}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-03/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-03/seed-731/checkpoint.json", "checkpoint_sha256": "7313f5fd95facd846729e367869e2528beff11428b29fee09f7e4c9e6b4be3d0", "completed_utc": "2026-08-02T00:48:06.431251+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.35038252221094, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 309.5413865899937, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4783296512, "peak_system_memory_used_bytes": 59830636544, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 154.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 143.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 148.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 155.99, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.841041226987727}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-03/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-04/seed-1731/checkpoint.json", "checkpoint_sha256": "c9e12feedf40562c1b1d1b5b5509199ca15bf10a743e43978476c873411ead0d", "completed_utc": "2026-08-02T00:52:16.576103+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.34822170770577, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 310.1338506556281, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4989853696, "peak_system_memory_used_bytes": 60058972160, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 159.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 142.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.71, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.829089127946645}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-04/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-04/seed-2731/checkpoint.json", "checkpoint_sha256": "3bc6ecb0620ad8500d4fc2667168246238dd577df963618d8f90f206657f311e", "completed_utc": "2026-08-02T00:53:18.983895+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.405424675595484, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 308.8006904731757, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5041475584, "peak_system_memory_used_bytes": 58865041408, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 167.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 148.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 145.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 147.13, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 57, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8444774539675564}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-04/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-04/seed-731/checkpoint.json", "checkpoint_sha256": "c52cd2fc59f3834e25e4a907056e69055288a67c9a3ec8273705a20012879590", "completed_utc": "2026-08-02T00:51:14.051832+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.359498060742986, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 306.3646944813335, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 4938227712, "peak_system_memory_used_bytes": 58645733376, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 151.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 143.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 149.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 148.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.01, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 57, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8411487430566922}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-04/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-05/seed-1731/checkpoint.json", "checkpoint_sha256": "67ba2a3d2f7387ea674f7c3f05b6a2e85da3ce14090a3a6caddaf731aa3fd541", "completed_utc": "2026-08-02T00:55:23.943225+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.40295380950494, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 307.8413238980182, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5144727552, "peak_system_memory_used_bytes": 60239429632, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 169.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 146.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 147.24, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 147.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 144.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 146.69, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 57, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8322917229961604}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-05/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-05/seed-2731/checkpoint.json", "checkpoint_sha256": "ad6140f2f96b3b89ee150fd9526860707bcb84dc5f8bf63ca02f13bb65a63e25", "completed_utc": "2026-08-02T00:56:26.371107+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.42550972555845, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 307.34134306880975, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5196353536, "peak_system_memory_used_bytes": 59018186752, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 149.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 142.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 156.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 149.17, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8500567820155993}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-05/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-05/seed-731/checkpoint.json", "checkpoint_sha256": "028305ff378c2f2992afcac1b74608bb5bb405c97d8ac72edd66674334064069", "completed_utc": "2026-08-02T00:54:21.450451+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.38091907221582, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 305.57619011697557, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5093097472, "peak_system_memory_used_bytes": 59399643136, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 157.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.28, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8383191160392016}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-05/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-06/seed-1731/checkpoint.json", "checkpoint_sha256": "f47c8c3413717e57e6574184550d8aaa242242c061bcf7cfaec3a04ba81e0366", "completed_utc": "2026-08-02T00:58:31.235255+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.384121862868554, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 310.38961038961037, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5299605504, "peak_system_memory_used_bytes": 60648706048, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 138.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 144.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 143.37, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8295170209603384}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-06/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-06/seed-2731/checkpoint.json", "checkpoint_sha256": "debaa865086377f748b49f78bafdcff82e3dba5a0a8bba30f4fdd0e4042170c2", "completed_utc": "2026-08-02T00:59:33.683482+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.40193070009452, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 308.78353499008404, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5351231488, "peak_system_memory_used_bytes": 59123564544, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 146.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 150.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 161.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 151.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 151.64, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.839739287039265}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-06/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-06/seed-731/checkpoint.json", "checkpoint_sha256": "b1a42003f15a13602226cf427867139914e03419e399f21b8244b4fa513a9af1", "completed_utc": "2026-08-02T00:57:28.758762+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.39443743572529, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 309.8809227450004, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5247979520, "peak_system_memory_used_bytes": 59114852352, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 149.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 147.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 143.66, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8279801370808855}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-06/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-07/seed-1731/checkpoint.json", "checkpoint_sha256": "fd205c5c5b458fe6d91a79c19cb39339dea6b5fc1bad052ea274ba6bb31ecedc", "completed_utc": "2026-08-02T01:01:38.718892+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.420117393057865, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 306.1417936066322, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5454508032, "peak_system_memory_used_bytes": 59499118592, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 150.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 145.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 151.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 154.51, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8519440459785983}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-07/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-07/seed-2731/checkpoint.json", "checkpoint_sha256": "f441bba2f4cac8ef0b4a1884868be43103c9b57ee3a9c32ed16ffe0ddf39efe2", "completed_utc": "2026-08-02T01:02:41.231657+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.32055015156864, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 309.6356275303644, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5506134016, "peak_system_memory_used_bytes": 59529666560, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 164.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 146.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 143.08, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8283122859429568}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-07/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-07/seed-731/checkpoint.json", "checkpoint_sha256": "fbf682a2faba1ab5e35c3812bc536306c08b396879b145707a016c845790c3b2", "completed_utc": "2026-08-02T01:00:36.282458+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.34290205157834, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 309.8367328948085, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5402853376, "peak_system_memory_used_bytes": 61239734272, "per_gpu": {"0": {"peak_memory_used_mib": 20086.0, "peak_power_draw_w": 160.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18520.0, "peak_power_draw_w": 144.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 18106.0, "peak_power_draw_w": 141.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22284.0, "peak_power_draw_w": 145.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19672.0, "peak_power_draw_w": 142.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 143.81, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8279455659212545}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-07/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-08", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-08/seed-1731/checkpoint.json", "checkpoint_sha256": "55138fcdd03426ccff16d838ff04f904de6ec00530e2ccde4639da2cf350a381", "completed_utc": "2026-08-02T01:04:46.229559+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.46558175215949, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 306.8894761306758, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5609381888, "peak_system_memory_used_bytes": 59630813184, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 160.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 146.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 143.23, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8467621100135148}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-08/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-08", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-08/seed-2731/checkpoint.json", "checkpoint_sha256": "6db78bf1c057d1dd4d625e9ddf13cf8c2ad91810c639c9c2d8225fd2ce3d452a", "completed_utc": "2026-08-02T01:05:48.619742+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.47552312410843, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 308.21409918303925, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5661003776, "peak_system_memory_used_bytes": 59580534784, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 145.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 163.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 152.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 149.67, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8420939210336655}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-08/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-INV-08", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-INV-08/seed-731/checkpoint.json", "checkpoint_sha256": "240fcd674d4814a4fcbbb36f123177b9050e0b25c71883503e19fc3e911a9f70", "completed_utc": "2026-08-02T01:03:43.771618+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 2048, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.3856402199416, "finish_reason": "length", "output_token_cap": 2048, "output_tokens": 2048, "prompt_tokens_per_second": 308.92961704360823, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5557764096, "peak_system_memory_used_bytes": 60986961920, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 166.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 149.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 151.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 151.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 152.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 154.24, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.8165828409837559}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "passed": false, "score": 0.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/R-INV-08/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "R-SCHED-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-01/seed-1731/checkpoint.json", "checkpoint_sha256": "93c6c88c5e7c22c848af54bc92f5b514b495916c675d3925a10cf3adb7eed408", "completed_utc": "2026-08-02T01:06:08.669854+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 266, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.5868014945139, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 266, "prompt_tokens_per_second": 299.6360303513673, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5752188928, "peak_system_memory_used_bytes": 59710722048, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 150.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 141.67, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7370312860002741}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 189, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "f79dcc7d0c667e80055a13354cb2fa0a186e8972febc22d8e3431278678df2b4", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":2},\"B\":{\"start\":0,\"finish\":1},\"C\":{\"start\":2,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":0,\"finish\":9},\"F\":{\"start\":9,\"finish\":10}},\"makespan\":12}"}]}
{"case_id": "R-SCHED-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-01/seed-2731/checkpoint.json", "checkpoint_sha256": "b27170ab487939cf65d73de934acc47adb21c9a9acf19d9657441c095d6343b0", "completed_utc": "2026-08-02T01:06:24.941368+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 521, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.44809165175955, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 521, "prompt_tokens_per_second": 299.0084352624752, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5791264768, "peak_system_memory_used_bytes": 59605065728, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 149.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 142.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 151.37, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7368465639883652}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 189, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "f79dcc7d0c667e80055a13354cb2fa0a186e8972febc22d8e3431278678df2b4", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":2},\"B\":{\"start\":0,\"finish\":1},\"C\":{\"start\":2,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":0,\"finish\":9},\"F\":{\"start\":9,\"finish\":10}},\"makespan\":12}"}]}
{"case_id": "R-SCHED-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-01/seed-731/checkpoint.json", "checkpoint_sha256": "a9f17d2a08ea8c58b4f4c128b5de5a4ae347bfc9af5a61a361880f758ec19b09", "completed_utc": "2026-08-02T01:05:59.857368+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 348, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.49307715020972, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 348, "prompt_tokens_per_second": 298.95190977514125, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5712584704, "peak_system_memory_used_bytes": 59786915840, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 154.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 145.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 142.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 146.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 143.95, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7507175579667091}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 189, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "f79dcc7d0c667e80055a13354cb2fa0a186e8972febc22d8e3431278678df2b4", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":2},\"B\":{\"start\":0,\"finish\":1},\"C\":{\"start\":2,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":0,\"finish\":9},\"F\":{\"start\":9,\"finish\":10}},\"makespan\":12}"}]}
{"case_id": "R-SCHED-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-02/seed-1731/checkpoint.json", "checkpoint_sha256": "6c16cc703b4ef7c026f109b85588b8862789749382ed6396e7e5dcb771c1d12a", "completed_utc": "2026-08-02T01:06:41.727327+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 268, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.590915315631136, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 268, "prompt_tokens_per_second": 297.761362602933, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5870891008, "peak_system_memory_used_bytes": 59692105728, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 140.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 143.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.51, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7464423120254651}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-02/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "491ec9d971baf545105f6664d53cff3257eb5b20c6f2c9961563249a470697ab", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":1},\"B\":{\"start\":1,\"finish\":8},\"C\":{\"start\":8,\"finish\":11},\"D\":{\"start\":0,\"finish\":6},\"E\":{\"start\":11,\"finish\":15},\"F\":{\"start\":8,\"finish\":10}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-02/seed-2731/checkpoint.json", "checkpoint_sha256": "51954eeff9937673ba9a919a9a611ff61962cadbfd0bc82cf1820d6545ea8770", "completed_utc": "2026-08-02T01:06:57.713356+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 511, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.457634473278645, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 511, "prompt_tokens_per_second": 294.1057157924628, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5909819392, "peak_system_memory_used_bytes": 59804000256, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 160.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 138.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 142.45, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 14, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7400594269856811}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-02/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "491ec9d971baf545105f6664d53cff3257eb5b20c6f2c9961563249a470697ab", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":1},\"B\":{\"start\":1,\"finish\":8},\"C\":{\"start\":8,\"finish\":11},\"D\":{\"start\":0,\"finish\":6},\"E\":{\"start\":11,\"finish\":15},\"F\":{\"start\":8,\"finish\":10}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-02/seed-731/checkpoint.json", "checkpoint_sha256": "751cb930358d046dd62523d9c8be5ea8cde5f8ca124e0c8d886c6855a3e7e35d", "completed_utc": "2026-08-02T01:06:32.888403+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 235, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.540187066713564, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 235, "prompt_tokens_per_second": 298.10723930668775, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5832065024, "peak_system_memory_used_bytes": 59671896064, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 161.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 141.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 142.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 148.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 146.79, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7378674719948322}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-02/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "491ec9d971baf545105f6664d53cff3257eb5b20c6f2c9961563249a470697ab", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":1},\"B\":{\"start\":1,\"finish\":8},\"C\":{\"start\":8,\"finish\":11},\"D\":{\"start\":0,\"finish\":6},\"E\":{\"start\":11,\"finish\":15},\"F\":{\"start\":8,\"finish\":10}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-03/seed-1731/checkpoint.json", "checkpoint_sha256": "4dfd30f82e702d06318ae53f6cc08ababaf1876dcbbfed8d4518c763f88b256f", "completed_utc": "2026-08-02T01:07:18.971594+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 204, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.57614225845259, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 204, "prompt_tokens_per_second": 297.50858651019485, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5990780928, "peak_system_memory_used_bytes": 59828510720, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 149.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 149.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 144.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 155.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 148.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 147.92, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7193496280815452}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-03/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fe44bbc412b9a92ffeebd24c82f6f4a272c74e8704837401cc8287b693af26a2", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":3},\"B\":{\"start\":3,\"finish\":11},\"C\":{\"start\":3,\"finish\":5},\"D\":{\"start\":5,\"finish\":13},\"E\":{\"start\":11,\"finish\":18},\"F\":{\"start\":0,\"finish\":9}},\"makespan\":18}"}]}
{"case_id": "R-SCHED-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-03/seed-2731/checkpoint.json", "checkpoint_sha256": "745e0872415ce575e3b1ab417833c303baaf8f8afcaee533108df8cca35813a4", "completed_utc": "2026-08-02T01:07:32.918450+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 447, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.45097060407349, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 447, "prompt_tokens_per_second": 297.13572924757, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6029312000, "peak_system_memory_used_bytes": 59977641984, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 163.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 145.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 146.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 151.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.11, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7177501219557598}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-03/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "fe44bbc412b9a92ffeebd24c82f6f4a272c74e8704837401cc8287b693af26a2", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":3},\"B\":{\"start\":3,\"finish\":11},\"C\":{\"start\":3,\"finish\":5},\"D\":{\"start\":5,\"finish\":13},\"E\":{\"start\":11,\"finish\":18},\"F\":{\"start\":0,\"finish\":9}},\"makespan\":18}"}]}
{"case_id": "R-SCHED-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-03/seed-731/checkpoint.json", "checkpoint_sha256": "d8cfa5ac935722bf4b5cee1f703708e111ea8d43945b9bc7c0344d349dd7a23f", "completed_utc": "2026-08-02T01:07:12.142553+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 463, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.44724924578752, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 463, "prompt_tokens_per_second": 296.97890126627686, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 5950455808, "peak_system_memory_used_bytes": 59858837504, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 156.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 151.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 148.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 157.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 145.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 145.42, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7308938639471307}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 257, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-03/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "4f5c723dffda018efdb07ea927e96afeeb084f435f97b75cc043624fb604da0c", "utf8": "{\n  \"tasks\": {\n    \"A\": {\"start\": 0, \"finish\": 3},\n    \"B\": {\"start\": 3, \"finish\": 11},\n    \"C\": {\"start\": 3, \"finish\": 5},\n    \"D\": {\"start\": 5, \"finish\": 13},\n    \"E\": {\"start\": 11, \"finish\": 18},\n    \"F\": {\"start\": 0, \"finish\": 9}\n  },\n  \"makespan\": 18\n}"}]}
{"case_id": "R-SCHED-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-04/seed-1731/checkpoint.json", "checkpoint_sha256": "620470937dc0e2742be4d672d97cf9af3197305b74da85b869c73bcbc44c7ebe", "completed_utc": "2026-08-02T01:07:53.373461+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 336, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.455682762426946, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 336, "prompt_tokens_per_second": 296.01761523542876, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6108704768, "peak_system_memory_used_bytes": 59988525056, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 148.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 143.25, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7278379620984197}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 223, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-04/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e59938afa77c3b51b2332d16404301173be34b6dbd0d7a6eb7b4bd5e9a0d9516", "utf8": "{\"tasks\": {\"A\": {\"start\": 0, \"finish\": 9}, \"B\": {\"start\": 0, \"finish\": 6}, \"C\": {\"start\": 0, \"finish\": 7}, \"D\": {\"start\": 7, \"finish\": 11}, \"E\": {\"start\": 6, \"finish\": 12}, \"F\": {\"start\": 11, \"finish\": 15}}, \"makespan\": 15}"}]}
{"case_id": "R-SCHED-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-04/seed-2731/checkpoint.json", "checkpoint_sha256": "749de107ba14b5f30c7eb92eaba875d6a29ebe99c9f381424acbad340c38cefb", "completed_utc": "2026-08-02T01:08:08.592590+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 487, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.41452657660048, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 487, "prompt_tokens_per_second": 299.45066292181235, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6148132864, "peak_system_memory_used_bytes": 60130103296, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.6, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7269326120149344}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-04/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1d34bbff60da03139ba72e62fc0c67dd2d52d5227a9ed2d46b87cf7215f35872", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":9},\"B\":{\"start\":0,\"finish\":6},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":11},\"E\":{\"start\":6,\"finish\":12},\"F\":{\"start\":11,\"finish\":15}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-04/seed-731/checkpoint.json", "checkpoint_sha256": "d455fa4a29a46aed1d1c41b7486d3158c69dac9a6307ba23a38ecc38b68cc424", "completed_utc": "2026-08-02T01:07:42.658194+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 304, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.60846554943558, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 304, "prompt_tokens_per_second": 300.12315398387614, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6069514240, "peak_system_memory_used_bytes": 59950120960, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 145.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 142.06, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 151.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 160.99, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7241030259756371}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-04/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1d34bbff60da03139ba72e62fc0c67dd2d52d5227a9ed2d46b87cf7215f35872", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":9},\"B\":{\"start\":0,\"finish\":6},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":11},\"E\":{\"start\":6,\"finish\":12},\"F\":{\"start\":11,\"finish\":15}},\"makespan\":15}"}]}
{"case_id": "R-SCHED-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-05/seed-1731/checkpoint.json", "checkpoint_sha256": "3dc2baa098c5b05a59b8dd44f15d4eae419d6bdfa8f44cf90eaaae6899958dbc", "completed_utc": "2026-08-02T01:08:26.051461+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 258, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.59364136691544, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 258, "prompt_tokens_per_second": 299.25334080238224, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6227632128, "peak_system_memory_used_bytes": 60196003840, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 165.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 138.62, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 144.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 143.51, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7169115910073742}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 190, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-05/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "2d4838c5698321ade1f862aa56e5197b4568702edd8457fc104de0df2aa72371", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":3},\"B\":{\"start\":0,\"finish\":8},\"C\":{\"start\":0,\"finish\":4},\"D\":{\"start\":8,\"finish\":13},\"E\":{\"start\":0,\"finish\":8},\"F\":{\"start\":13,\"finish\":20}},\"makespan\":20}"}]}
{"case_id": "R-SCHED-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-05/seed-2731/checkpoint.json", "checkpoint_sha256": "23edb7eb4e836562398ac89a8ea7ca4dc2296d27a73879d7c5a0a7617449a1fd", "completed_utc": "2026-08-02T01:08:33.016493+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 208, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.478140196074534, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 208, "prompt_tokens_per_second": 299.4568472357035, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6266507264, "peak_system_memory_used_bytes": 60113129472, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 164.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 149.36, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 138.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 153.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 160.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 141.67, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7180280199972913}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 190, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-05/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "2d4838c5698321ade1f862aa56e5197b4568702edd8457fc104de0df2aa72371", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":3},\"B\":{\"start\":0,\"finish\":8},\"C\":{\"start\":0,\"finish\":4},\"D\":{\"start\":8,\"finish\":13},\"E\":{\"start\":0,\"finish\":8},\"F\":{\"start\":13,\"finish\":20}},\"makespan\":20}"}]}
{"case_id": "R-SCHED-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-05/seed-731/checkpoint.json", "checkpoint_sha256": "aa1f4f840fe8dccd80a1dd9239e14f95ece35a134cf4656cf734f550ad261c7c", "completed_utc": "2026-08-02T01:08:17.654419+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 278, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.566611165288954, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 278, "prompt_tokens_per_second": 297.4585609453028, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6188597248, "peak_system_memory_used_bytes": 60154322944, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 146.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 148.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 151.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 145.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 141.0, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7375424149213359}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 190, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-05/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "2d4838c5698321ade1f862aa56e5197b4568702edd8457fc104de0df2aa72371", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":3},\"B\":{\"start\":0,\"finish\":8},\"C\":{\"start\":0,\"finish\":4},\"D\":{\"start\":8,\"finish\":13},\"E\":{\"start\":0,\"finish\":8},\"F\":{\"start\":13,\"finish\":20}},\"makespan\":20}"}]}
{"case_id": "R-SCHED-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-06/seed-1731/checkpoint.json", "checkpoint_sha256": "2f8ba4921cccada9a1be5d88fe0cbe7e6189e8368858bfb0d24c8e63276b7873", "completed_utc": "2026-08-02T01:08:47.933529+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 225, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.430429644324484, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 225, "prompt_tokens_per_second": 296.7947634824898, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6343720960, "peak_system_memory_used_bytes": 60218003456, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 154.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 141.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 151.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 142.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 149.66, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7210444789379835}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-06/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8f2df7d48aa132b0a78bc8a0a5d48922df0d7c7ef15551d86c790ff2acbff5fc", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":7},\"B\":{\"start\":0,\"finish\":2},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":2,\"finish\":10},\"F\":{\"start\":10,\"finish\":18}},\"makespan\":18}"}]}
{"case_id": "R-SCHED-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-06/seed-2731/checkpoint.json", "checkpoint_sha256": "5211bcb6be22c253f1e0aeca170332c59c63e74f5c4dfc84dc905b3e01ec389e", "completed_utc": "2026-08-02T01:09:00.742550+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 407, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.32011679970561, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 407, "prompt_tokens_per_second": 296.5124557249656, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6382354432, "peak_system_memory_used_bytes": 61790728192, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 149.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 145.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 144.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 148.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 147.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 150.12, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 11, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7112102350220084}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-06/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8f2df7d48aa132b0a78bc8a0a5d48922df0d7c7ef15551d86c790ff2acbff5fc", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":7},\"B\":{\"start\":0,\"finish\":2},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":2,\"finish\":10},\"F\":{\"start\":10,\"finish\":18}},\"makespan\":18}"}]}
{"case_id": "R-SCHED-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-06/seed-731/checkpoint.json", "checkpoint_sha256": "ecd8cd0d782fe0204a5b61a4561018ccfe6f3f0d6745dc350e0715e32953c4bb", "completed_utc": "2026-08-02T01:08:40.496676+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 227, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.40697755316068, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 227, "prompt_tokens_per_second": 297.88369853579843, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6305034240, "peak_system_memory_used_bytes": 60138242048, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 145.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 138.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 144.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 145.23, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7071661129593849}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-06/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "8f2df7d48aa132b0a78bc8a0a5d48922df0d7c7ef15551d86c790ff2acbff5fc", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":7},\"B\":{\"start\":0,\"finish\":2},\"C\":{\"start\":0,\"finish\":7},\"D\":{\"start\":7,\"finish\":12},\"E\":{\"start\":2,\"finish\":10},\"F\":{\"start\":10,\"finish\":18}},\"makespan\":18}"}]}
{"case_id": "R-SCHED-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-07/seed-1731/checkpoint.json", "checkpoint_sha256": "3a7c31557802784a4e04aec6fd6928ac105de4fe9f0eb9b44074150f4a470ee4", "completed_utc": "2026-08-02T01:09:24.095134+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 413, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.412035380238414, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 413, "prompt_tokens_per_second": 298.611456726223, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6461640704, "peak_system_memory_used_bytes": 60591075328, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 147.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 144.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 144.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 151.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 148.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 151.4, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 12, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7269106699386612}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 1731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-07/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "6abc19a4f7733697224a9ff2a40ce7c618de36f1b51b1785ea8c07344fa7d46f", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":9},\"B\":{\"start\":0,\"finish\":9},\"C\":{\"start\":0,\"finish\":4},\"D\":{\"start\":9,\"finish\":10},\"E\":{\"start\":9,\"finish\":10},\"F\":{\"start\":10,\"finish\":13}},\"makespan\":13}"}]}
{"case_id": "R-SCHED-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-07/seed-2731/checkpoint.json", "checkpoint_sha256": "5c0227282e03664d89dae1f5de54d7ed24ec317466325178a8e97bb78f45788e", "completed_utc": "2026-08-02T01:09:39.044240+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 476, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.393764381615284, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 476, "prompt_tokens_per_second": 297.2547648190243, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6501634048, "peak_system_memory_used_bytes": 60587802624, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 156.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 145.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 139.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 145.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 141.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 145.15, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 13, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7426376959774643}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 2731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 191, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-07/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "6abc19a4f7733697224a9ff2a40ce7c618de36f1b51b1785ea8c07344fa7d46f", "utf8": "{\"tasks\":{\"A\":{\"start\":0,\"finish\":9},\"B\":{\"start\":0,\"finish\":9},\"C\":{\"start\":0,\"finish\":4},\"D\":{\"start\":9,\"finish\":10},\"E\":{\"start\":9,\"finish\":10},\"F\":{\"start\":10,\"finish\":13}},\"makespan\":13}"}]}
{"case_id": "R-SCHED-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-07/seed-731/checkpoint.json", "checkpoint_sha256": "e2b09d3cef17bc8d5ff1cc80a45b74a67e8107c028d324f04c2d2b5e9d82031d", "completed_utc": "2026-08-02T01:09:11.083387+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 322, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.283274606367854, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 322, "prompt_tokens_per_second": 296.72511472643833, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6422282240, "peak_system_memory_used_bytes": 61811085312, "per_gpu": {"0": {"peak_memory_used_mib": 19819.0, "peak_power_draw_w": 144.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18253.0, "peak_power_draw_w": 143.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17839.0, "peak_power_draw_w": 138.47, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22017.0, "peak_power_draw_w": 142.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19405.0, "peak_power_draw_w": 143.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16835.0, "peak_power_draw_w": 144.82, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 0.7450041749980301}, "quant": "UD-IQ3_XXS", "score": {"max_score": 1.0, "parse_error": null, "passed": true, "score": 1.0, "scorer": "reasoning_exact_json"}, "seed": 731, "status": "completed", "task": "reasoning", "visible_model_outputs": [{"bytes": 223, "path": "results/quants/UD-IQ3_XXS/cases/R-SCHED-07/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "4425e2339ccff33b11d4fd0c7da4fd638c12402debea13ee33c2270971058599", "utf8": "{\"tasks\": {\"A\": {\"start\": 0, \"finish\": 9}, \"B\": {\"start\": 0, \"finish\": 9}, \"C\": {\"start\": 0, \"finish\": 4}, \"D\": {\"start\": 9, \"finish\": 10}, \"E\": {\"start\": 9, \"finish\": 10}, \"F\": {\"start\": 10, \"finish\": 13}}, \"makespan\": 13}"}]}
{"case_id": "STRICT-JSON-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/STRICT-JSON-01/seed-1731/checkpoint.json", "checkpoint_sha256": "7bfd962430c80ce80fd0965e4786b12bc1e8513812817922f33cbeef5607985d", "completed_utc": "2026-08-02T01:13:55.692375+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": [989, 361, 582, 803, 4628, 17462, 6288, 3362, 5062, 201, 223, 7413, 223, 875, 361, 582, 86, 29929, 5165, 3362], "id": "repeated_20_token_ngram", "passed": false}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4064, "passed": true, "task_completion_valid": true}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.351726183905406, "finish_reason": "stop", "output_token_cap": 4096, "output_tokens": 4064, "prompt_tokens_per_second": 714.0475179946471, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6615564288, "peak_system_memory_used_bytes": 62112587776, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 181.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 169.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 153.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 173.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 165.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 149.41, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 115, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.9395060440292582}, "quant": "UD-IQ3_XXS", "score": {"components": {"exact_shape": 1.0, "factual_slots": 6.0, "parse": 1.0, "schema_validity": 2.0}, "factual_slots_correct": 72, "factual_slots_total": 72, "max_score": 10.0, "parse_error": null, "passed": true, "score": 10.0, "scorer": "strict_json_ajv_equivalent"}, "seed": 1731, "status": "completed", "task": "strict_json", "visible_model_outputs": [{"bytes": 2184, "path": "results/quants/UD-IQ3_XXS/cases/STRICT-JSON-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9a4b44f10dbeb47ff6300d9ad5e7d61929b997496b1ec630e147227c0a28e94e", "utf8": "[\n  {\n    \"ticket_id\": \"T-4107\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-07-31T22:14:00-07:00\",\n    \"customer\": \"Alma\",\n    \"refund_cents\": 1840,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4108\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T09:02:11+02:00\",\n    \"customer\": \"Benoît\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4109\",\n    \"priority\": \"low\",\n    \"opened_at\": \"2026-08-01T17:33:20+09:00\",\n    \"customer\": \"Chika\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4110\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T08:05:00-03:00\",\n    \"customer\": \"Diego\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4111\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T12:00:00Z\",\n    \"customer\": \"Eleni\",\n    \"refund_cents\": 10299,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4112\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T15:45:30+04:00\",\n    \"customer\": \"Faisal\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4113\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-08-01T07:18:44-05:00\",\n    \"customer\": \"Grace\",\n    \"refund_cents\": 2400,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4114\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-02T00:03:09+09:00\",\n    \"customer\": \"Haruto\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4115\",\n    \"priority\": \"low\",\n    \"opened_at\": \"2026-08-01T13:27:00+01:00\",\n    \"customer\": \"Inês\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4116\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-08-01T10:11:12-04:00\",\n    \"customer\": \"Jamal\",\n    \"refund_cents\": 5550,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4117\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T16:40:05+02:00\",\n    \"customer\": \"Klara\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4118\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T18:22:31Z\",\n    \"customer\": \"Luis\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  }\n]"}]}
{"case_id": "STRICT-JSON-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/STRICT-JSON-01/seed-2731/checkpoint.json", "checkpoint_sha256": "fca24e99962582a96b78dbf831c5dc05da225bf24748e6924090332b9e4fbf89", "completed_utc": "2026-08-02T01:14:57.654793+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 4, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": [989, 361, 582, 803, 4628, 17462, 6288, 3362, 5062, 201, 223, 7413, 223, 875, 361, 582, 86, 29929, 5165, 3362], "id": "repeated_20_token_ngram", "passed": false}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 1946, "passed": true, "task_completion_valid": true}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.573609955928035, "finish_reason": "stop", "output_token_cap": 4096, "output_tokens": 1946, "prompt_tokens_per_second": 714.5720693877108, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6688571392, "peak_system_memory_used_bytes": 60618821632, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 158.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 155.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 166.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 164.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 166.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 160.44, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 56, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.9221978810383007}, "quant": "UD-IQ3_XXS", "score": {"components": {"exact_shape": 1.0, "factual_slots": 6.0, "parse": 1.0, "schema_validity": 2.0}, "factual_slots_correct": 72, "factual_slots_total": 72, "max_score": 10.0, "parse_error": null, "passed": true, "score": 10.0, "scorer": "strict_json_ajv_equivalent"}, "seed": 2731, "status": "completed", "task": "strict_json", "visible_model_outputs": [{"bytes": 2184, "path": "results/quants/UD-IQ3_XXS/cases/STRICT-JSON-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9a4b44f10dbeb47ff6300d9ad5e7d61929b997496b1ec630e147227c0a28e94e", "utf8": "[\n  {\n    \"ticket_id\": \"T-4107\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-07-31T22:14:00-07:00\",\n    \"customer\": \"Alma\",\n    \"refund_cents\": 1840,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4108\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T09:02:11+02:00\",\n    \"customer\": \"Benoît\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4109\",\n    \"priority\": \"low\",\n    \"opened_at\": \"2026-08-01T17:33:20+09:00\",\n    \"customer\": \"Chika\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4110\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T08:05:00-03:00\",\n    \"customer\": \"Diego\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4111\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T12:00:00Z\",\n    \"customer\": \"Eleni\",\n    \"refund_cents\": 10299,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4112\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T15:45:30+04:00\",\n    \"customer\": \"Faisal\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4113\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-08-01T07:18:44-05:00\",\n    \"customer\": \"Grace\",\n    \"refund_cents\": 2400,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4114\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-02T00:03:09+09:00\",\n    \"customer\": \"Haruto\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4115\",\n    \"priority\": \"low\",\n    \"opened_at\": \"2026-08-01T13:27:00+01:00\",\n    \"customer\": \"Inês\",\n    \"refund_cents\": null,\n    \"needs_human\": false\n  },\n  {\n    \"ticket_id\": \"T-4116\",\n    \"priority\": \"high\",\n    \"opened_at\": \"2026-08-01T10:11:12-04:00\",\n    \"customer\": \"Jamal\",\n    \"refund_cents\": 5550,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4117\",\n    \"priority\": \"urgent\",\n    \"opened_at\": \"2026-08-01T16:40:05+02:00\",\n    \"customer\": \"Klara\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  },\n  {\n    \"ticket_id\": \"T-4118\",\n    \"priority\": \"medium\",\n    \"opened_at\": \"2026-08-01T18:22:31Z\",\n    \"customer\": \"Luis\",\n    \"refund_cents\": null,\n    \"needs_human\": true\n  }\n]"}]}
{"case_id": "STRICT-JSON-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/STRICT-JSON-01/seed-731/checkpoint.json", "checkpoint_sha256": "e31b05c81c84a2d732fd3ab7ce700ba3ed4106ef38062898b9701799cda73e83", "completed_utc": "2026-08-02T01:11:47.809456+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.3789726019189, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 715.6704703729246, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6542323712, "peak_system_memory_used_bytes": 60434780160, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 179.81, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 167.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 152.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 149.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 177.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 161.38, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 116, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.9147384960670024}, "quant": "UD-IQ3_XXS", "score": {"components": {"exact_shape": 0.0, "factual_slots": 0.0, "parse": 0.0, "schema_validity": 0.0}, "max_score": 10.0, "parse_error": "JSON parse failure at line 1, column 1: Expecting value", "score": 0.0, "scorer": "strict_json_ajv_equivalent"}, "seed": 731, "status": "completed", "task": "strict_json", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/STRICT-JSON-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "TC-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-01/seed-1731/checkpoint.json", "checkpoint_sha256": "57d925f8ab11b7a81ef97fecfcbaffaf6a911d9a5acd8c11f3d214af94445bab", "completed_utc": "2026-08-02T01:15:17.731777+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 372, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.87234581212942, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 372, "prompt_tokens_per_second": 607.4043465913446, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6746939392, "peak_system_memory_used_bytes": 62703099904, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 151.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 150.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 154.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 153.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 151.65, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 11, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3935472049051896}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"amount_cents": 1840, "invoice_id": "INV-8841", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 121, "path": "results/quants/UD-IQ3_XXS/cases/TC-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ea3c45ab80c1244f2974490a94b75690071f54a20bb2ef2af0d0659d6a2e9e65", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-8841\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-01/seed-2731/checkpoint.json", "checkpoint_sha256": "35cdb1e825137fea659a08c33295680e60ddb1a93c18eaa398788786a5522c1a", "completed_utc": "2026-08-02T01:15:23.466294+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 136, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.27899395643682, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 136, "prompt_tokens_per_second": 602.2190904899328, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6747103232, "peak_system_memory_used_bytes": 62061158400, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 158.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 159.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 154.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 159.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 157.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 148.51, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3900412059156224}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"amount_cents": 1840, "invoice_id": "INV-8841", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 121, "path": "results/quants/UD-IQ3_XXS/cases/TC-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ea3c45ab80c1244f2974490a94b75690071f54a20bb2ef2af0d0659d6a2e9e65", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-8841\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-01/seed-731/checkpoint.json", "checkpoint_sha256": "981ce525124be1eb12d06b61e6658e736d7e98703c3afdf328f53aded71079eb", "completed_utc": "2026-08-02T01:15:04.697472+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 176, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.58859392420505, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 176, "prompt_tokens_per_second": 609.4023754973368, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6746902528, "peak_system_memory_used_bytes": 60932395008, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 182.06, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 154.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 152.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 150.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 151.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 152.21, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4174821510678157}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"amount_cents": 1840, "invoice_id": "INV-8841", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 121, "path": "results/quants/UD-IQ3_XXS/cases/TC-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ea3c45ab80c1244f2974490a94b75690071f54a20bb2ef2af0d0659d6a2e9e65", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-8841\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-02/seed-1731/checkpoint.json", "checkpoint_sha256": "08861f13036a1f5b59e44ad21a24b0e3cb59a97cc3b023f9ccff045d2c5531b7", "completed_utc": "2026-08-02T01:15:51.724410+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 117, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.141593634774544, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 117, "prompt_tokens_per_second": 640.4975753067857, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6747828224, "peak_system_memory_used_bytes": 60770406400, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 168.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 157.61, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 149.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 150.98, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 155.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 146.22, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.5187118789181113}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "lee@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"amount_cents": 1840, "invoice_id": "INV-9920", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ3_XXS/cases/TC-02/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcc872a46c28b70c804a2802d00324e1ecc112dd9defb810ff2369d74e4038a2", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"lee@example.test\"}}"}, {"bytes": 121, "path": "results/quants/UD-IQ3_XXS/cases/TC-02/seed-1731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "e5f900c628fee389011b5077910c06588f7a30f594d669fad58825ed813c431d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-9920\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-02/seed-2731/checkpoint.json", "checkpoint_sha256": "498086807b71f556cacdef452e88077656ea2d07ed76d7882c028fb2258a6c0a", "completed_utc": "2026-08-02T01:16:02.239714+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 125, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.11953130299125, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 125, "prompt_tokens_per_second": 644.7102896030569, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6748176384, "peak_system_memory_used_bytes": 60789411840, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 169.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 161.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 152.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 167.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 160.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 164.65, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.5289781510364264}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "lee@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"amount_cents": 1840, "invoice_id": "INV-9920", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ3_XXS/cases/TC-02/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcc872a46c28b70c804a2802d00324e1ecc112dd9defb810ff2369d74e4038a2", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"lee@example.test\"}}"}, {"bytes": 121, "path": "results/quants/UD-IQ3_XXS/cases/TC-02/seed-2731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "e5f900c628fee389011b5077910c06588f7a30f594d669fad58825ed813c431d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-9920\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-02", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-02/seed-731/checkpoint.json", "checkpoint_sha256": "01ff54f7f1cdd357cc2ac9d7eb3fee84ae98979e381c51c6c10a35ea59f3c942", "completed_utc": "2026-08-02T01:15:39.493386+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 110, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.202585575891305, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 110, "prompt_tokens_per_second": 645.962699519124, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6747492352, "peak_system_memory_used_bytes": 60769435648, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 163.86, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 161.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 140.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 157.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 160.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 155.24, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.5277538960799575}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "lee@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"amount_cents": 1840, "invoice_id": "INV-9920", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ3_XXS/cases/TC-02/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcc872a46c28b70c804a2802d00324e1ecc112dd9defb810ff2369d74e4038a2", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"lee@example.test\"}}"}, {"bytes": 121, "path": "results/quants/UD-IQ3_XXS/cases/TC-02/seed-731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "e5f900c628fee389011b5077910c06588f7a30f594d669fad58825ed813c431d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-9920\",\"amount_cents\":1840,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-03/seed-1731/checkpoint.json", "checkpoint_sha256": "06f664f457e6b5f473e8fcf3fa67c41ff47f427dcdefaf3a82745c472c97adfb", "completed_utc": "2026-08-02T01:16:11.569591+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 92, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.50797544448148, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 92, "prompt_tokens_per_second": 608.0950839585826, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6748512256, "peak_system_memory_used_bytes": 60789424128, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 151.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 150.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 152.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 171.26, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 159.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 161.65, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4069239790551364}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-117", "note": "Deceased-owner access request.", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 135, "path": "results/quants/UD-IQ3_XXS/cases/TC-03/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b3016a3eb67c2409660d39dc9bf8f3c733ba57dd6feba22859434bde5d41bf03", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-117\",\"queue\":\"legal\",\"note\":\"Deceased-owner access request.\"}}"}]}
{"case_id": "TC-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-03/seed-2731/checkpoint.json", "checkpoint_sha256": "9cf75eae8d8357fb6157c2b3afe489ce4437f76a4b62f18e44a662521ebf87fb", "completed_utc": "2026-08-02T01:16:16.937488+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 122, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.671557690418176, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 122, "prompt_tokens_per_second": 609.8228158349685, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6748692480, "peak_system_memory_used_bytes": 60808679424, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 174.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 160.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 159.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 160.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 159.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 151.52, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3945118659175932}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-117", "note": "Deceased-owner access request.", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 135, "path": "results/quants/UD-IQ3_XXS/cases/TC-03/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "b3016a3eb67c2409660d39dc9bf8f3c733ba57dd6feba22859434bde5d41bf03", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-117\",\"queue\":\"legal\",\"note\":\"Deceased-owner access request.\"}}"}]}
{"case_id": "TC-03", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-03/seed-731/checkpoint.json", "checkpoint_sha256": "62e9e90d9f80269aa0709c4c7d76c841b6aad2e9d5b86c45f2bd2bbe7c48c3dd", "completed_utc": "2026-08-02T01:16:06.994689+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 101, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.61020152809946, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 101, "prompt_tokens_per_second": 611.7926696114677, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6748344320, "peak_system_memory_used_bytes": 60806213632, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 150.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 149.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 150.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 153.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 153.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 151.63, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3899313749279827}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-117", "note": "Deceased-owner access request", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 134, "path": "results/quants/UD-IQ3_XXS/cases/TC-03/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ec1f0c9ec982ed3ce05d30e2c9ca33489dad55589409145652dff83da55907bc", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-117\",\"queue\":\"legal\",\"note\":\"Deceased-owner access request\"}}"}]}
{"case_id": "TC-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-04/seed-1731/checkpoint.json", "checkpoint_sha256": "058f9cea516b76cf4ede02e56b889b3dcfc386f4f41333c324ee8d3c5cb61aac", "completed_utc": "2026-08-02T01:16:36.229722+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 233, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.35263873767277, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 233, "prompt_tokens_per_second": 607.3660534394209, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6749052928, "peak_system_memory_used_bytes": 60797108224, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 153.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 154.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 148.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 148.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 151.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 153.09, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 8, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3954901889665052}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"kind": "ask_user", "message": "Which invoice should I refund $9.00 against?"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 76, "path": "results/quants/UD-IQ3_XXS/cases/TC-04/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "9579cc9694a8f6e188dd24320115095f1988544c11d2c2c593b796e4c852de6f", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Which invoice should I refund $9.00 against?\"}"}]}
{"case_id": "TC-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-04/seed-2731/checkpoint.json", "checkpoint_sha256": "06a4d6755ad01d24716079ad0ce59c8281d70b3aaba33c3ff9f4315d33ad3517", "completed_utc": "2026-08-02T01:16:41.317008+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 116, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.65560509441413, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 116, "prompt_tokens_per_second": 609.326741726313, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6749241344, "peak_system_memory_used_bytes": 60791070720, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 166.75, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 149.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 150.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 154.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 154.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 141.16, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3838555829133838}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"kind": "ask_user", "message": "Which invoice should the $9.00 refund be issued against? Please provide an invoice ID (or customer/email if you can)."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 149, "path": "results/quants/UD-IQ3_XXS/cases/TC-04/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "1052bff3981144b0ea6039799bc2d739121a372d2a5bed8f2d2cc33ec117811d", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Which invoice should the $9.00 refund be issued against? Please provide an invoice ID (or customer/email if you can).\"}"}]}
{"case_id": "TC-04", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-04/seed-731/checkpoint.json", "checkpoint_sha256": "e917cce775acf729fd9b22374d56928e5f227234532432ea8c1a97f91bba1130", "completed_utc": "2026-08-02T01:16:27.567429+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 295, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.38150076437979, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 295, "prompt_tokens_per_second": 603.3992705346816, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6748872704, "peak_system_memory_used_bytes": 60780449792, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 176.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 158.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 140.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 158.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 158.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 162.37, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3981567339506}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"kind": "ask_user", "message": "Please provide the invoice identifier and the refund reason (duplicate, fraud, or service_failure)."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 131, "path": "results/quants/UD-IQ3_XXS/cases/TC-04/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "481482eeacc99c9f0239e851e0ebaaa0d980a1861936d954498860fba08ab346", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Please provide the invoice identifier and the refund reason (duplicate, fraud, or service_failure).\"}"}]}
{"case_id": "TC-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-05/seed-1731/checkpoint.json", "checkpoint_sha256": "f1622c47aec435f1509560aa0061b8d33dc799694a7b3f6c4e0a53933280f35f", "completed_utc": "2026-08-02T01:16:50.497802+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 71, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.06687471450999, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 71, "prompt_tokens_per_second": 602.3959861712106, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6749589504, "peak_system_memory_used_bytes": 60759261184, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 156.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 149.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 151.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 165.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 147.77, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3990805939538404}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-3310"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-05/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "620e7956cad16843a07c4a9447ddc6b4480d9c8e414c5524ebcfcef65dad591d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-3310\"}}"}]}
{"case_id": "TC-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-05/seed-2731/checkpoint.json", "checkpoint_sha256": "8d295707c5833f476fbd25ddc34ba2d09debd11dcfc0290a1dd2aa583fa59065", "completed_utc": "2026-08-02T01:16:53.637448+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 49, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.23640587608517, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 49, "prompt_tokens_per_second": 605.9514135802242, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6749749248, "peak_system_memory_used_bytes": 60661538816, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 181.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 150.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 154.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 151.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 153.54, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 150.62, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.38303286395967}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-3310"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-05/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "620e7956cad16843a07c4a9447ddc6b4480d9c8e414c5524ebcfcef65dad591d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-3310\"}}"}]}
{"case_id": "TC-05", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-05/seed-731/checkpoint.json", "checkpoint_sha256": "236df3240ba796041563cd5a95e8e3e168b75ae95ae2c075a8b44fcfe43d32a5", "completed_utc": "2026-08-02T01:16:46.776931+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 127, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.73906103927465, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 127, "prompt_tokens_per_second": 607.3636951299603, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6749417472, "peak_system_memory_used_bytes": 60760117248, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 154.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 152.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 140.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 160.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 164.88, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 141.09, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3803604179993272}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-3310"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-05/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "620e7956cad16843a07c4a9447ddc6b4480d9c8e414c5524ebcfcef65dad591d", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-3310\"}}"}]}
{"case_id": "TC-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-06/seed-1731/checkpoint.json", "checkpoint_sha256": "547ad9b7e0b520773cfe9a95b9756bfadef58c828a2763a1e265a557613f9745", "completed_utc": "2026-08-02T01:17:38.855481+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 473, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.089397548404435, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 473, "prompt_tokens_per_second": 633.8627474060333, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6750429184, "peak_system_memory_used_bytes": 60854611968, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 174.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 148.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 158.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 161.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 150.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 155.34, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 15, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.5094468609895557}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "amy@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-606"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ3_XXS/cases/TC-06/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "5f1f41cbae0966614e1c06469d63fedbfcad8f31107fe71236878b61cc785043", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"amy@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ3_XXS/cases/TC-06/seed-1731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "2519b5e57415d9a165fbf2aac2692445aa6f524f1b49f31c3f864a483728e054", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-606\"}}"}]}
{"case_id": "TC-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-06/seed-2731/checkpoint.json", "checkpoint_sha256": "ca455f80fb5b1452cc4e4ad22dd2597f9433eb60a8614c0adc6d057ea636a6db", "completed_utc": "2026-08-02T01:17:52.340876+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 262, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.416835968335434, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 262, "prompt_tokens_per_second": 632.1611477941489, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6750781440, "peak_system_memory_used_bytes": 60759326720, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 171.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 168.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 154.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 168.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 157.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 157.95, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 9, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.492977779940702}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "amy@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-606"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ3_XXS/cases/TC-06/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "5f1f41cbae0966614e1c06469d63fedbfcad8f31107fe71236878b61cc785043", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"amy@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ3_XXS/cases/TC-06/seed-2731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "2519b5e57415d9a165fbf2aac2692445aa6f524f1b49f31c3f864a483728e054", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-606\"}}"}]}
{"case_id": "TC-06", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-06/seed-731/checkpoint.json", "checkpoint_sha256": "9ccfcfa6d590ff018bcc5f9c8e560c4caf9196980acad08af2408a4997559592", "completed_utc": "2026-08-02T01:17:19.684579+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 665, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.01474517953692, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 665, "prompt_tokens_per_second": 637.1276084323873, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6750089216, "peak_system_memory_used_bytes": 60763119616, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 166.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 155.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 149.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 157.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 165.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 160.54, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 19, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.5012452400987968}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "amy@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-606"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 87, "path": "results/quants/UD-IQ3_XXS/cases/TC-06/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "5f1f41cbae0966614e1c06469d63fedbfcad8f31107fe71236878b61cc785043", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"amy@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ3_XXS/cases/TC-06/seed-731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "2519b5e57415d9a165fbf2aac2692445aa6f524f1b49f31c3f864a483728e054", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-606\"}}"}]}
{"case_id": "TC-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-07/seed-1731/checkpoint.json", "checkpoint_sha256": "93d36e799ce5555f546a3b4b0415066a227845dbcbec69d6859f1d1dcd75b28f", "completed_utc": "2026-08-02T01:18:00.613025+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 67, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.585525540789625, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 67, "prompt_tokens_per_second": 604.9050291745253, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6751125504, "peak_system_memory_used_bytes": 60650196992, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 153.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 151.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 165.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 140.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 165.9, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3684376369928941}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-808"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 78, "path": "results/quants/UD-IQ3_XXS/cases/TC-07/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ae97c01bf7ef4a65020b0039d0f8cc2ecaf3668fff9be418d02cd00c64383426", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-808\"}}"}]}
{"case_id": "TC-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-07/seed-2731/checkpoint.json", "checkpoint_sha256": "2ac2e97fecf98c48ce113b18656232b0e4d345c09a39df901f737488d3d5b1df", "completed_utc": "2026-08-02T01:18:03.527028+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 43, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.232071151757935, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 43, "prompt_tokens_per_second": 601.0533789509938, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6751297536, "peak_system_memory_used_bytes": 60860776448, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 157.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 155.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 154.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 169.85, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 161.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 151.79, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 2, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3962962919613346}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-808"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 78, "path": "results/quants/UD-IQ3_XXS/cases/TC-07/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ae97c01bf7ef4a65020b0039d0f8cc2ecaf3668fff9be418d02cd00c64383426", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-808\"}}"}]}
{"case_id": "TC-07", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-07/seed-731/checkpoint.json", "checkpoint_sha256": "475897c508aa60b7660086b520d36bdede21d7b57c62eb4a90a03f69dcac485d", "completed_utc": "2026-08-02T01:17:57.053895+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 104, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.03682780379428, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 104, "prompt_tokens_per_second": 600.3106717271932, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6750945280, "peak_system_memory_used_bytes": 59959341056, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 167.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 157.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 154.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 159.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 152.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 141.6, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3796372599899769}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-808"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 78, "path": "results/quants/UD-IQ3_XXS/cases/TC-07/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ae97c01bf7ef4a65020b0039d0f8cc2ecaf3668fff9be418d02cd00c64383426", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-808\"}}"}]}
{"case_id": "TC-08", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-08/seed-1731/checkpoint.json", "checkpoint_sha256": "02db41888351be1436af1629b746d4298220faf9cdfc0ed9d75a9fccfc75ab2d", "completed_utc": "2026-08-02T01:18:14.992695+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 115, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 31.95090562758247, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 115, "prompt_tokens_per_second": 594.6357819927337, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6751641600, "peak_system_memory_used_bytes": 62124867584, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 173.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 143.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 167.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 160.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 144.22, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4298157460289076}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"amount_cents": 2200, "invoice_id": "INV-7001", "reason": "fraud"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 117, "path": "results/quants/UD-IQ3_XXS/cases/TC-08/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e6d61ba2e18b0b4002df102f492835cd369c3caa7ce725f57fa20c1809daf77e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-7001\",\"amount_cents\":2200,\"reason\":\"fraud\"}}"}]}
{"case_id": "TC-08", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-08/seed-2731/checkpoint.json", "checkpoint_sha256": "a230595ebf7c5ee4a3d8a4a29e49f70050973c5ef63b59deb4d239abbd5b75a0", "completed_utc": "2026-08-02T01:18:22.151427+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 184, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.182682751371, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 184, "prompt_tokens_per_second": 597.070030257317, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6751805440, "peak_system_memory_used_bytes": 62026350592, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 150.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 151.36, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 153.04, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 163.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 156.7, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3992642150260508}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"amount_cents": 2200, "invoice_id": "INV-7001", "reason": "fraud"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 117, "path": "results/quants/UD-IQ3_XXS/cases/TC-08/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e6d61ba2e18b0b4002df102f492835cd369c3caa7ce725f57fa20c1809daf77e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-7001\",\"amount_cents\":2200,\"reason\":\"fraud\"}}"}]}
{"case_id": "TC-08", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-08/seed-731/checkpoint.json", "checkpoint_sha256": "46a60e0b4d05de9d21c899292e07152fe0bf31604e8822faff2f62f7cf2099af", "completed_utc": "2026-08-02T01:18:09.632620+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 146, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.44419549644129, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 146, "prompt_tokens_per_second": 601.8845966246486, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6751453184, "peak_system_memory_used_bytes": 62074261504, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 171.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 153.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 147.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 148.99, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 148.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 168.12, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3993396039586514}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"amount_cents": 2200, "invoice_id": "INV-7001", "reason": "fraud"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 117, "path": "results/quants/UD-IQ3_XXS/cases/TC-08/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e6d61ba2e18b0b4002df102f492835cd369c3caa7ce725f57fa20c1809daf77e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-7001\",\"amount_cents\":2200,\"reason\":\"fraud\"}}"}]}
{"case_id": "TC-09", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-09/seed-1731/checkpoint.json", "checkpoint_sha256": "4c0843fa266cf03ac6ed02c6dfff8f284fc420af44dfb8f44c2b1ad925425e83", "completed_utc": "2026-08-02T01:18:34.264398+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 124, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.672136578530754, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 124, "prompt_tokens_per_second": 606.4028533851031, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6752133120, "peak_system_memory_used_bytes": 60805550080, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 163.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 156.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 155.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 166.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 150.29, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 151.45, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.391562337987125}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-7440"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-09/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d6c84e227bb66da4c0cdd7cab54902164e489bc38b561066cbda2d965f3bc42e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-7440\"}}"}]}
{"case_id": "TC-09", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-09/seed-2731/checkpoint.json", "checkpoint_sha256": "f24be0a9992e3c0e8d54110cccc822619a4a36bf8521073e4f1234964d5ce1ce", "completed_utc": "2026-08-02T01:18:45.291866+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 308, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.209960486616495, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 308, "prompt_tokens_per_second": 604.9293349133981, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6752309248, "peak_system_memory_used_bytes": 60803596288, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 158.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 141.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 154.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 167.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 154.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 153.86, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 10, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3943867880152538}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"amount_cents": 6500, "invoice_id": "INV-7440", "reason": "service_failure"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 127, "path": "results/quants/UD-IQ3_XXS/cases/TC-09/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "322e392693a24fe4431bb53ad571a7dc8da4c0d4a3e509f5b3678bf957db1739", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-7440\",\"amount_cents\":6500,\"reason\":\"service_failure\"}}"}]}
{"case_id": "TC-09", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-09/seed-731/checkpoint.json", "checkpoint_sha256": "a325c7f7b4f4a38f78ec56f3a5e5f13a54374da48d0e92577ea27a5824982cc1", "completed_utc": "2026-08-02T01:18:28.768095+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 164, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.74376589069036, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 164, "prompt_tokens_per_second": 608.5980441171181, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6751969280, "peak_system_memory_used_bytes": 60942348288, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 176.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 152.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 147.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 150.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 150.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 154.57, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3853853619657457}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"amount_cents": 6500, "invoice_id": "INV-7440", "reason": "service_failure"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 127, "path": "results/quants/UD-IQ3_XXS/cases/TC-09/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "322e392693a24fe4431bb53ad571a7dc8da4c0d4a3e509f5b3678bf957db1739", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-7440\",\"amount_cents\":6500,\"reason\":\"service_failure\"}}"}]}
{"case_id": "TC-10", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-10/seed-1731/checkpoint.json", "checkpoint_sha256": "eb195cd87e35b6c2ca3f3a9080f4b4d3997b8653d391952fa12aa756be727e4a", "completed_utc": "2026-08-02T01:18:55.563816+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 134, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.65194695348171, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 134, "prompt_tokens_per_second": 597.5317398192556, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6752555008, "peak_system_memory_used_bytes": 60595617792, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 163.93, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 150.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 152.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 152.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 154.28, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 140.51, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4174172210041434}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-1010"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-10/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "317f118bb6adea7882e5c29b2af7050ccc4ae34b5e6174d8a9a813feb2be56f7", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-1010\"}}"}]}
{"case_id": "TC-10", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-10/seed-2731/checkpoint.json", "checkpoint_sha256": "69a2943430c6be1304675bd8c140ebafb83e295356b9ffa8eb3f5ec849f985ea", "completed_utc": "2026-08-02T01:19:00.114146+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 91, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.27606924055234, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 91, "prompt_tokens_per_second": 595.5750711960067, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6752608256, "peak_system_memory_used_bytes": 60816232448, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 150.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 149.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 153.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 165.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 150.18, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 136.63, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4167139020282775}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-1010"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-10/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "317f118bb6adea7882e5c29b2af7050ccc4ae34b5e6174d8a9a813feb2be56f7", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-1010\"}}"}]}
{"case_id": "TC-10", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-10/seed-731/checkpoint.json", "checkpoint_sha256": "856f7a695ba254c2c3bfec2e7fdd390d4294c3bc16403187a6d5ee90a12fa520", "completed_utc": "2026-08-02T01:18:49.839200+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 91, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.58227269322077, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 91, "prompt_tokens_per_second": 601.7541605602048, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6752501760, "peak_system_memory_used_bytes": 60665114624, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 176.21, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 159.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 149.67, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 150.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 156.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 159.51, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4083391909953207}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-1010"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-10/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "317f118bb6adea7882e5c29b2af7050ccc4ae34b5e6174d8a9a813feb2be56f7", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-1010\"}}"}]}
{"case_id": "TC-11", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-11/seed-1731/checkpoint.json", "checkpoint_sha256": "0979e5bc7357685ed9c2eb64a95390ce74de0672b0d73cda7ac96b815dc8a4d9", "completed_utc": "2026-08-02T01:19:11.252729+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 91, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.55757908986683, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 91, "prompt_tokens_per_second": 607.8925809506657, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6752972800, "peak_system_memory_used_bytes": 60663091200, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 148.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 149.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 153.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 166.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 151.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 152.85, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4020568890264258}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-1111"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-11/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "bd32332a1d2ba332e0ccf2c95409f8527ca6deb74504b01f6ba3f6f4394f93f6", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-1111\"}}"}]}
{"case_id": "TC-11", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-11/seed-2731/checkpoint.json", "checkpoint_sha256": "dbf0022c7c8b12951e1f51f82016ca9bcf550e35ca39d895fa0f35ae098e6fc6", "completed_utc": "2026-08-02T01:19:16.326684+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 111, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.708643139046025, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 111, "prompt_tokens_per_second": 604.966799104042, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6753157120, "peak_system_memory_used_bytes": 60572319744, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 153.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 150.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 152.08, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 166.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 149.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 169.93, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4079945670673624}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-1111"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-11/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "bd32332a1d2ba332e0ccf2c95409f8527ca6deb74504b01f6ba3f6f4394f93f6", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-1111\"}}"}]}
{"case_id": "TC-11", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-11/seed-731/checkpoint.json", "checkpoint_sha256": "539c4ab1ae08c343f78b95a526d724dc1b6616e4d3e6f791415200148d3fb8a8", "completed_utc": "2026-08-02T01:19:06.729591+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 162, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.72830848160776, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 162, "prompt_tokens_per_second": 608.607764293557, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6752796672, "peak_system_memory_used_bytes": 60706643968, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 150.22, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 150.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 158.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 152.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 145.85, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4068109189392999}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"kind": "ask_user", "message": "Which amount should I refund: $9.00 or $19.00?"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 78, "path": "results/quants/UD-IQ3_XXS/cases/TC-11/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "74fe91c90ee54a990f01cd52e51c2fe6f830bdef31def9ce4ac49487b7bdea60", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Which amount should I refund: $9.00 or $19.00?\"}"}]}
{"case_id": "TC-12", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-12/seed-1731/checkpoint.json", "checkpoint_sha256": "6e23745fd4e041b4fd03ed922bc0d5061a7a79a771bcad10dd13a0bc1ec29b1d", "completed_utc": "2026-08-02T01:19:24.682216+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 60, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.15586345168909, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 60, "prompt_tokens_per_second": 609.2904669193495, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6753165312, "peak_system_memory_used_bytes": 61842092032, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 150.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 149.17, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 150.82, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 152.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 153.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 151.74, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3897627129917964}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-515", "note": "Stolen security key.", "queue": "security"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 128, "path": "results/quants/UD-IQ3_XXS/cases/TC-12/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "61f7c1032c41cca6ad19c9628767cbd10bcc53cda75048231b8ea84627011f3e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-515\",\"queue\":\"security\",\"note\":\"Stolen security key.\"}}"}]}
{"case_id": "TC-12", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-12/seed-2731/checkpoint.json", "checkpoint_sha256": "8aeb404aba57daf07a1843de27c8a00aef6b1ec0b0bc1ba8336b213473def0c8", "completed_utc": "2026-08-02T01:19:28.410648+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 68, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.76660522468696, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 68, "prompt_tokens_per_second": 608.5243321647317, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6753349632, "peak_system_memory_used_bytes": 61773991936, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 153.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 153.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.57, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 161.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 147.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 168.21, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 2, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3873207170981914}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-515", "note": "Stolen security key", "queue": "security"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 127, "path": "results/quants/UD-IQ3_XXS/cases/TC-12/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "530a64689607898b4b9710807e46435d2acefd9c3bc1b562285ddeae3b3976b5", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-515\",\"queue\":\"security\",\"note\":\"Stolen security key\"}}"}]}
{"case_id": "TC-12", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-12/seed-731/checkpoint.json", "checkpoint_sha256": "28d10f64bfd3a3f13d002becf8d31bfcd7252f497cded86b4cbfe6e78b619759", "completed_utc": "2026-08-02T01:19:21.151027+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 103, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.67986535901592, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 103, "prompt_tokens_per_second": 606.0632645243171, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6753157120, "peak_system_memory_used_bytes": 61847019520, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 158.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 143.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 155.66, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 160.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 156.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 141.35, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.391992894001305}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-515", "note": "Stolen security key.", "queue": "security"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 128, "path": "results/quants/UD-IQ3_XXS/cases/TC-12/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "61f7c1032c41cca6ad19c9628767cbd10bcc53cda75048231b8ea84627011f3e", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-515\",\"queue\":\"security\",\"note\":\"Stolen security key.\"}}"}]}
{"case_id": "TC-13", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-13/seed-1731/checkpoint.json", "checkpoint_sha256": "ea4633405098d8d2ce5d9f94e871c1c77f23a41fcdd6dfb757ccf11833613164", "completed_utc": "2026-08-02T01:19:37.268444+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 63, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.11939731063328, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 63, "prompt_tokens_per_second": 603.6190901098324, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6753767424, "peak_system_memory_used_bytes": 65066369024, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 153.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 153.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 161.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 160.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 166.03, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.396285191993229}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-616", "note": "Repeated invoice mismatch", "queue": "billing"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 132, "path": "results/quants/UD-IQ3_XXS/cases/TC-13/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "60c790db56251d37104cdee67d0e98f5f6abdec0e6c07276196b272bb73c0c87", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-616\",\"queue\":\"billing\",\"note\":\"Repeated invoice mismatch\"}}"}]}
{"case_id": "TC-13", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-13/seed-2731/checkpoint.json", "checkpoint_sha256": "4e7551cb5938464ab139735dbd5464c748d34aa7f2d126cb73f563e76983a516", "completed_utc": "2026-08-02T01:19:41.065919+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 69, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.920385414234275, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 69, "prompt_tokens_per_second": 602.9079385791177, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6753931264, "peak_system_memory_used_bytes": 66244128768, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 154.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 151.34, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 147.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 173.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 155.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 164.55, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.397254224983044}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-616", "note": "Repeated invoice mismatch.", "queue": "billing"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 133, "path": "results/quants/UD-IQ3_XXS/cases/TC-13/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "7c1e6c5631edd80383226c77059e81122ffb16a425f5b71361e5ec9e81927b4f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-616\",\"queue\":\"billing\",\"note\":\"Repeated invoice mismatch.\"}}"}]}
{"case_id": "TC-13", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-13/seed-731/checkpoint.json", "checkpoint_sha256": "24da158ce1002b1cb4249449eb08587e8afd574fb587e2687c0692792ea44a68", "completed_utc": "2026-08-02T01:19:33.665031+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 119, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.79521067905938, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 119, "prompt_tokens_per_second": 609.0004082067953, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6753583104, "peak_system_memory_used_bytes": 61713227776, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 162.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 155.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 149.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 158.7, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 158.32, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 164.18, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3836539880139753}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"customer_id": "C-616", "note": "Repeated invoice mismatch.", "queue": "billing"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 133, "path": "results/quants/UD-IQ3_XXS/cases/TC-13/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "7c1e6c5631edd80383226c77059e81122ffb16a425f5b71361e5ec9e81927b4f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-616\",\"queue\":\"billing\",\"note\":\"Repeated invoice mismatch.\"}}"}]}
{"case_id": "TC-14", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-14/seed-1731/checkpoint.json", "checkpoint_sha256": "38d11cfc839b29b138ed9c21deadbdc8c9babe08bd1eb498c9a968e7da80111f", "completed_utc": "2026-08-02T01:19:59.732777+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 96, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.174705254838756, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 96, "prompt_tokens_per_second": 640.5907234075368, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6754525184, "peak_system_memory_used_bytes": 61865938944, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 171.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 158.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 153.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 160.42, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 153.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 158.5, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.5171442410210148}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "owner@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-714", "note": "Estate access review", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 89, "path": "results/quants/UD-IQ3_XXS/cases/TC-14/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d69329385f42792b3f62ce14173757205b66eb3a684ab367f8fd6912e0cea5e6", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"owner@example.test\"}}"}, {"bytes": 125, "path": "results/quants/UD-IQ3_XXS/cases/TC-14/seed-1731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "6a11d3f7a4a318fccc827137a8c22d0f8140c674678aa4eb3206a8e766bdebec", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-714\",\"queue\":\"legal\",\"note\":\"Estate access review\"}}"}]}
{"case_id": "TC-14", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-14/seed-2731/checkpoint.json", "checkpoint_sha256": "8b637035b0d3d93f209b78d166893538a13bff35c1e03a6eee39bb11e528cdda", "completed_utc": "2026-08-02T01:20:08.649042+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 116, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.3320690134742, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 116, "prompt_tokens_per_second": 640.9622962558868, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6754738176, "peak_system_memory_used_bytes": 62710063104, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 159.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 165.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 149.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 163.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 151.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 176.0, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.5082545350305736}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "owner@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-714", "note": "Estate access review", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 89, "path": "results/quants/UD-IQ3_XXS/cases/TC-14/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d69329385f42792b3f62ce14173757205b66eb3a684ab367f8fd6912e0cea5e6", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"owner@example.test\"}}"}, {"bytes": 125, "path": "results/quants/UD-IQ3_XXS/cases/TC-14/seed-2731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "6a11d3f7a4a318fccc827137a8c22d0f8140c674678aa4eb3206a8e766bdebec", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-714\",\"queue\":\"legal\",\"note\":\"Estate access review\"}}"}]}
{"case_id": "TC-14", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-14/seed-731/checkpoint.json", "checkpoint_sha256": "39075e054d85ebca2e247edce511fa8eb9a7719011731c3a51c2f4a639fd0298", "completed_utc": "2026-08-02T01:19:49.356595+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 72, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.12154092878835, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 72, "prompt_tokens_per_second": 641.0816204539561, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6754312192, "peak_system_memory_used_bytes": 62796619776, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 197.13, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 148.95, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 166.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 162.0, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 154.41, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4955876249587163}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "owner@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-714", "note": "Estate access review", "queue": "legal"}, "kind": "tool_call", "name": "escalate_case"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 89, "path": "results/quants/UD-IQ3_XXS/cases/TC-14/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "d69329385f42792b3f62ce14173757205b66eb3a684ab367f8fd6912e0cea5e6", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"owner@example.test\"}}"}, {"bytes": 125, "path": "results/quants/UD-IQ3_XXS/cases/TC-14/seed-731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "6a11d3f7a4a318fccc827137a8c22d0f8140c674678aa4eb3206a8e766bdebec", "utf8": "{\"kind\":\"tool_call\",\"name\":\"escalate_case\",\"arguments\":{\"customer_id\":\"C-714\",\"queue\":\"legal\",\"note\":\"Estate access review\"}}"}]}
{"case_id": "TC-15", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-15/seed-1731/checkpoint.json", "checkpoint_sha256": "7540a0b75327db3c683e9bd973f1b296eecbecacd980101766f910026291a6d1", "completed_utc": "2026-08-02T01:20:16.337392+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 65, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.16731365541371, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 65, "prompt_tokens_per_second": 601.5355294759745, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6755098624, "peak_system_memory_used_bytes": 64104833024, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 179.45, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 157.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 149.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 156.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 155.48, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 159.18, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4016924460884184}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8080"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-15/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ad3a40d83ab15f4a3e7b92cb50eae47c0b668ecd18b6c8c66f8c6d0c83a85d7c", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-8080\"}}"}]}
{"case_id": "TC-15", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-15/seed-2731/checkpoint.json", "checkpoint_sha256": "1affa11f5e039e31319d42e8ac6c9c054eb1d1c910a482531c484232646a2d13", "completed_utc": "2026-08-02T01:20:19.597178+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 52, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.26555409048347, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 52, "prompt_tokens_per_second": 601.5359707506802, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6755225600, "peak_system_memory_used_bytes": 63138058240, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 145.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 150.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 150.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 153.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 154.09, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 141.46, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3939838770311326}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8080"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-15/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ad3a40d83ab15f4a3e7b92cb50eae47c0b668ecd18b6c8c66f8c6d0c83a85d7c", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-8080\"}}"}]}
{"case_id": "TC-15", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-15/seed-731/checkpoint.json", "checkpoint_sha256": "b11a099ea5fc52ab6a8e774fc5f18a4f477eebe33572cf5afa4c8ddc68cd521d", "completed_utc": "2026-08-02T01:20:12.685129+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 78, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.68102514677154, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 78, "prompt_tokens_per_second": 598.7215105305621, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6754918400, "peak_system_memory_used_bytes": 65432825856, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 145.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 153.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 162.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 149.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 153.67, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3937650579027832}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8080"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-15/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ad3a40d83ab15f4a3e7b92cb50eae47c0b668ecd18b6c8c66f8c6d0c83a85d7c", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-8080\"}}"}]}
{"case_id": "TC-16", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-16/seed-1731/checkpoint.json", "checkpoint_sha256": "be9e2462737b6c611aee159d0fdba18da813c7b447274c28707e7fc4692953ae", "completed_utc": "2026-08-02T01:20:26.788411+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 68, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.123612787322074, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 68, "prompt_tokens_per_second": 604.63470413111, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6755225600, "peak_system_memory_used_bytes": 62935035904, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 147.52, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 149.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 150.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 154.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 154.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 140.37, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.391591768944636}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"kind": "ask_user", "message": "Please provide the customer ID or the email address of the account to escalate."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 111, "path": "results/quants/UD-IQ3_XXS/cases/TC-16/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "23b4e16384b0efb421fe45e3b795efb44a08909fbda5b3c4bec1b528a6dd38ee", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Please provide the customer ID or the email address of the account to escalate.\"}"}]}
{"case_id": "TC-16", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-16/seed-2731/checkpoint.json", "checkpoint_sha256": "ce4ebcb1fca3917ee4bcb535a7e420cd017fe430d6ae9464c760eb8db579814d", "completed_utc": "2026-08-02T01:20:33.564873+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 168, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.75649511133616, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 168, "prompt_tokens_per_second": 604.8416587435091, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6755348480, "peak_system_memory_used_bytes": 62891122688, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 171.9, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 155.43, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 157.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 161.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 162.27, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3915058189304546}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"kind": "ask_user", "message": "Please provide the customer ID (C-xxxx) or the account email so I can look up the account before escalating to security."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 152, "path": "results/quants/UD-IQ3_XXS/cases/TC-16/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "4b4c76ccbe63b305ea727af5915fb9d436b9fc7d2945dfff724e42d53d692889", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Please provide the customer ID (C-xxxx) or the account email so I can look up the account before escalating to security.\"}"}]}
{"case_id": "TC-16", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-16/seed-731/checkpoint.json", "checkpoint_sha256": "a0d3eb83e4db6900dda8d99c16ab13fa439273e9f1c7d4ce39c944dcadf34bc0", "completed_utc": "2026-08-02T01:20:23.007531+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 55, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.20215087999025, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 55, "prompt_tokens_per_second": 605.3463723136382, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6755225600, "peak_system_memory_used_bytes": 63027929088, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 146.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 149.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 151.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 153.25, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 152.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 140.76, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.390205606003292}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"kind": "ask_user", "message": "Please provide the customer ID or email for the account to escalate."}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 100, "path": "results/quants/UD-IQ3_XXS/cases/TC-16/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "4db77f3822590c338d9aceac65e5ef0cded1cb3629ec32478a243bce3e05a1ab", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Please provide the customer ID or email for the account to escalate.\"}"}]}
{"case_id": "TC-17", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-17/seed-1731/checkpoint.json", "checkpoint_sha256": "7c3ea0ed3aac36b1902563484845687a0cfc618e2a3b5f0d281900a1d90eec12", "completed_utc": "2026-08-02T01:20:42.973042+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 115, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.564735429621315, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 115, "prompt_tokens_per_second": 607.0143554858097, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6755586048, "peak_system_memory_used_bytes": 61446389760, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 155.58, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 159.07, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 156.53, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 161.31, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 160.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 158.19, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3858047750545666}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-3250"}, "kind": "tool_call", "name": "get_invoice"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 2.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-17/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "eb276a823cf317bcb00410a85266c101c0e248ef9384e4bbf964c4a675ae2dde", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-3250\"}}"}]}
{"case_id": "TC-17", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-17/seed-2731/checkpoint.json", "checkpoint_sha256": "c367abb18f3471937aa3d86096276f78f50231fa6e549ad4dfc37ac38e11aa15", "completed_utc": "2026-08-02T01:20:52.125355+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 242, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.39545992862204, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 242, "prompt_tokens_per_second": 603.7677443470694, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6755692544, "peak_system_memory_used_bytes": 61301678080, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 171.3, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 155.27, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 150.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 148.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 149.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 167.57, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3939389809966087}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"amount_cents": 325, "invoice_id": "INV-3250", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 120, "path": "results/quants/UD-IQ3_XXS/cases/TC-17/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "68524c26479d8c763f3370fa7c6d957caa35a3984bbee1ad0b5eac104ca49eea", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-3250\",\"amount_cents\":325,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-17", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-17/seed-731/checkpoint.json", "checkpoint_sha256": "fc342ff4f65768ea043562a23629e3a646724a56ef81e7a77ae1869445dd0046", "completed_utc": "2026-08-02T01:20:37.781177+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 85, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.519134496513026, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 85, "prompt_tokens_per_second": 601.8616227703101, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6755479552, "peak_system_memory_used_bytes": 61503307776, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 172.41, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 155.2, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 155.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 154.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 155.38, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 159.37, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.397721122018993}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"amount_cents": 325, "invoice_id": "INV-3250", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 120, "path": "results/quants/UD-IQ3_XXS/cases/TC-17/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "68524c26479d8c763f3370fa7c6d957caa35a3984bbee1ad0b5eac104ca49eea", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-3250\",\"amount_cents\":325,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-18", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-18/seed-1731/checkpoint.json", "checkpoint_sha256": "681be073059e2211768eb121a225dca7505b2f3aa43ceeed8b34eb89e03de76a", "completed_utc": "2026-08-02T01:21:19.761711+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 198, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.29498244614537, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 198, "prompt_tokens_per_second": 638.5097454248059, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6756147200, "peak_system_memory_used_bytes": 60998672384, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 175.59, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 169.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 156.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 157.56, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 168.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 170.91, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.5020871930755675}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "zoë@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-918"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 88, "path": "results/quants/UD-IQ3_XXS/cases/TC-18/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcd2013a41e2c21a2375fd2fe8398c8d7a72305abece93dd5f039d3501fb6e9f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"zoë@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ3_XXS/cases/TC-18/seed-1731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "08453e209245616b2e7178a214ff1e085fab00d101b565907129428a4c41c828", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-918\"}}"}]}
{"case_id": "TC-18", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-18/seed-2731/checkpoint.json", "checkpoint_sha256": "9e7f58550dfd6d063f52cf4fd35b503d91cd5108e6c21c7197e3c8035f9c98ff", "completed_utc": "2026-08-02T01:21:28.746579+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 121, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.23018245292574, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 121, "prompt_tokens_per_second": 625.8455572954949, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6756315136, "peak_system_memory_used_bytes": 60965634048, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 183.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 154.83, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 161.1, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 176.11, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 152.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 150.08, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.531186220003292}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "zoë@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-918"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 88, "path": "results/quants/UD-IQ3_XXS/cases/TC-18/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcd2013a41e2c21a2375fd2fe8398c8d7a72305abece93dd5f039d3501fb6e9f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"zoë@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ3_XXS/cases/TC-18/seed-2731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "08453e209245616b2e7178a214ff1e085fab00d101b565907129428a4c41c828", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-918\"}}"}]}
{"case_id": "TC-18", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-18/seed-731/checkpoint.json", "checkpoint_sha256": "767306f4befea4872aaff741f5077f1d9a5cd89ebc5683486333d9f5109ea096", "completed_utc": "2026-08-02T01:21:06.326223+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 231, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.21926613895489, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 231, "prompt_tokens_per_second": 638.9025014392299, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6755958784, "peak_system_memory_used_bytes": 61174317056, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 156.91, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 152.36, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 163.33, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 148.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 165.4, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 156.61, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 7, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.5025480720214546}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"email": "zoë@example.test"}, "kind": "tool_call", "name": "search_customers"}, {"arguments": {"customer_id": "C-918"}, "kind": "tool_call", "name": "get_customer"}], "components": {"schema_valid_arguments": 0.0, "sequence_or_abstention": 0.0, "tool_choice": 0.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 0.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 88, "path": "results/quants/UD-IQ3_XXS/cases/TC-18/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "dcd2013a41e2c21a2375fd2fe8398c8d7a72305abece93dd5f039d3501fb6e9f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"search_customers\",\"arguments\":{\"email\":\"zoë@example.test\"}}"}, {"bytes": 78, "path": "results/quants/UD-IQ3_XXS/cases/TC-18/seed-731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "08453e209245616b2e7178a214ff1e085fab00d101b565907129428a4c41c828", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_customer\",\"arguments\":{\"customer_id\":\"C-918\"}}"}]}
{"case_id": "TC-19", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-19/seed-1731/checkpoint.json", "checkpoint_sha256": "9f3bfe2bdb307ff27380123ed1076cc37eff5e833b6eacc9bbab9ee0317a05b7", "completed_utc": "2026-08-02T01:22:30.636360+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 171, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.708516348236174, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 171, "prompt_tokens_per_second": 609.9953026709028, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6810001408, "peak_system_memory_used_bytes": 60967497728, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 167.01, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 154.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 147.64, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 155.15, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 159.46, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 164.81, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 6, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4337671319954097}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"kind": "ask_user", "message": "What type of issue is this: security, billing, or legal?"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 88, "path": "results/quants/UD-IQ3_XXS/cases/TC-19/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "ec2f0d9133eda3b9d5053229be2992eb35710381ba4ce6eafd8f00f3e232dab8", "utf8": "{\"kind\":\"ask_user\",\"message\":\"What type of issue is this: security, billing, or legal?\"}"}]}
{"case_id": "TC-19", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-19/seed-2731/checkpoint.json", "checkpoint_sha256": "59ab86a3e2c19015a1ce5a75f67387702700ac601963c346972f57e938f4b781", "completed_utc": "2026-08-02T01:22:36.874795+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 154, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.63552155698392, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 154, "prompt_tokens_per_second": 609.4712563136842, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6810132480, "peak_system_memory_used_bytes": 60990902272, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 155.02, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 158.87, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 140.69, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 165.92, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 157.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 160.87, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 5, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3997972370125353}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"kind": "ask_user", "message": "Which queue should the escalation for C-919 go to: security, billing, or legal?"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 111, "path": "results/quants/UD-IQ3_XXS/cases/TC-19/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "445b0315c52bf9635497faf3a1f603e729ab6d061ed8cba8489e8a7f25f3d295", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Which queue should the escalation for C-919 go to: security, billing, or legal?\"}"}]}
{"case_id": "TC-19", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-19/seed-731/checkpoint.json", "checkpoint_sha256": "659ae8ee6e707494fbe08976abb550491f97690eecc43dff466e3e4e7273a521", "completed_utc": "2026-08-02T01:22:23.821436+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 1, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 1738, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.57597930132022, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 1738, "prompt_tokens_per_second": 606.9879947224756, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6756446208, "peak_system_memory_used_bytes": 61062443008, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 162.23, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 154.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 144.76, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 169.14, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 155.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 155.56, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 49, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.3990289809880778}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"kind": "ask_user", "message": "Which queue (security, billing, or legal) should I use for C-919, and what note should be included in the escalation?"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 0.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": false, "score": 4.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 149, "path": "results/quants/UD-IQ3_XXS/cases/TC-19/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "388d8c841f7378a38115644b3f42ca9a8d83c2cce16f212d85f00f78e1f24401", "utf8": "{\"kind\":\"ask_user\",\"message\":\"Which queue (security, billing, or legal) should I use for C-919, and what note should be included in the escalation?\"}"}]}
{"case_id": "TC-20", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-20/seed-1731/checkpoint.json", "checkpoint_sha256": "1078daeb4dfced424217317fe61c503842c99f6c80172cfa6af95de2f68309ab", "completed_utc": "2026-08-02T01:22:51.747343+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 93, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.63695811945533, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 93, "prompt_tokens_per_second": 636.7839075512355, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6810284032, "peak_system_memory_used_bytes": 60942671872, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 163.05, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 163.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 137.49, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 154.39, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 158.97, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 149.33, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4977888650028035}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8820"}, "kind": "tool_call", "name": "get_invoice"}, {"arguments": {"amount_cents": 1275, "invoice_id": "INV-8820", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 1731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-20/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "18ba214e3084f670760bb3f93bae86f57a2254f07c8f8f9539bebe37154d9e29", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-8820\"}}"}, {"bytes": 121, "path": "results/quants/UD-IQ3_XXS/cases/TC-20/seed-1731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "8fcbfee0badae01da9ff392b97f3fc374979123ed9ef3185e5c3e5ed6d90529f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-8820\",\"amount_cents\":1275,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-20", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-20/seed-2731/checkpoint.json", "checkpoint_sha256": "8f0bc3d44dfbcc5d2fb17145c4a6319dbeda108ab3adac337384ade99fcc0943", "completed_utc": "2026-08-02T01:22:59.749310+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 85, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 33.828422242386615, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 85, "prompt_tokens_per_second": 635.3486229243297, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6810443776, "peak_system_memory_used_bytes": 64985952256, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 159.89, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 152.37, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 166.63, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 157.65, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 151.55, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 168.24, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 4, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.5032565459841862}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8820"}, "kind": "tool_call", "name": "get_invoice"}, {"arguments": {"amount_cents": 1275, "invoice_id": "INV-8820", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 2731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-20/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "18ba214e3084f670760bb3f93bae86f57a2254f07c8f8f9539bebe37154d9e29", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-8820\"}}"}, {"bytes": 121, "path": "results/quants/UD-IQ3_XXS/cases/TC-20/seed-2731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "8fcbfee0badae01da9ff392b97f3fc374979123ed9ef3185e5c3e5ed6d90529f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-8820\",\"amount_cents\":1275,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TC-20", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TC-20/seed-731/checkpoint.json", "checkpoint_sha256": "b7341c19bd68b4b630343f6f700d0453c9dc2ad68130a348796edd7eed87006e", "completed_utc": "2026-08-02T01:22:44.063406+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "stop", "id": "premature_eos", "passed": true, "task_completion_valid": true}, {"finish_reason": "stop", "id": "runaway_generation", "output_token_cap": 2048, "output_tokens": 64, "passed": true, "task_completion_valid": true}], "passed": true, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 34.18188071912268, "finish_reason": "stop", "output_token_cap": 2048, "output_tokens": 64, "prompt_tokens_per_second": 637.1388581108799, "request_count": 2, "telemetry": {"peak_process_rss_bytes": 6810251264, "peak_system_memory_used_bytes": 60969025536, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 164.19, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 153.94, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 170.44, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 163.16, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 164.35, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 160.53, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 3, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4914902310119942}, "quant": "UD-IQ3_XXS", "score": {"actual_actions": [{"arguments": {"invoice_id": "INV-8820"}, "kind": "tool_call", "name": "get_invoice"}, {"arguments": {"amount_cents": 1275, "invoice_id": "INV-8820", "reason": "duplicate"}, "kind": "tool_call", "name": "issue_refund"}], "components": {"schema_valid_arguments": 2.0, "sequence_or_abstention": 1.0, "tool_choice": 2.0}, "max_score": 5.0, "parse_errors": [], "passed": true, "score": 5.0, "scorer": "raw_tool_sequence"}, "seed": 731, "status": "completed", "task": "tool_calling", "visible_model_outputs": [{"bytes": 79, "path": "results/quants/UD-IQ3_XXS/cases/TC-20/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "18ba214e3084f670760bb3f93bae86f57a2254f07c8f8f9539bebe37154d9e29", "utf8": "{\"kind\":\"tool_call\",\"name\":\"get_invoice\",\"arguments\":{\"invoice_id\":\"INV-8820\"}}"}, {"bytes": 121, "path": "results/quants/UD-IQ3_XXS/cases/TC-20/seed-731/attempts/attempt-001/turn-02/chat/model_output.bin", "sha256": "8fcbfee0badae01da9ff392b97f3fc374979123ed9ef3185e5c3e5ed6d90529f", "utf8": "{\"kind\":\"tool_call\",\"name\":\"issue_refund\",\"arguments\":{\"invoice_id\":\"INV-8820\",\"amount_cents\":1275,\"reason\":\"duplicate\"}}"}]}
{"case_id": "TS-PATCH-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TS-PATCH-01/seed-1731/checkpoint.json", "checkpoint_sha256": "f84dcb013a92ab32b73690ee6c93223bb2b6078af457e2d053480665b393d5d8", "completed_utc": "2026-08-02T01:27:18.307642+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.162074492751046, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 580.4655829850661, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6922813440, "peak_system_memory_used_bytes": 65389547520, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 168.71, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 156.51, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 150.12, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 166.74, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 165.79, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 152.1, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 116, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4636037519667298}, "quant": "UD-IQ3_XXS", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "typescript_patch"}, "seed": 1731, "status": "completed", "task": "typescript_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/TS-PATCH-01/seed-1731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "TS-PATCH-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TS-PATCH-01/seed-2731/checkpoint.json", "checkpoint_sha256": "f8d04dfd2a9d2891abc2f374390b93bcbb686a1344a292c6b897d7e2cd8cce50", "completed_utc": "2026-08-02T01:29:27.537702+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.17910740341173, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 577.1993215732597, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6992371712, "peak_system_memory_used_bytes": 50813358080, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 153.8, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 150.78, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 152.96, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 156.72, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 158.77, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 160.25, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 116, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4697732779895887}, "quant": "UD-IQ3_XXS", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "typescript_patch"}, "seed": 2731, "status": "completed", "task": "typescript_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/TS-PATCH-01/seed-2731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}
{"case_id": "TS-PATCH-01", "checkpoint_path": "results/quants/UD-IQ3_XXS/cases/TS-PATCH-01/seed-731/checkpoint.json", "checkpoint_sha256": "c0d45e5021a3452d3f5c5fc8d3737247cb691f79471fb2e3d2c417069bf3cdcf", "completed_utc": "2026-08-02T01:25:09.133217+00:00", "corruption": {"checks": [{"error": null, "id": "invalid_utf8", "passed": true}, {"count": 0, "id": "replacement_character", "passed": true}, {"id": "unexpected_control_character", "occurrences": [], "passed": true}, {"id": "degenerate_spaces", "maximum_observed": 0, "passed": true}, {"id": "degenerate_blank_lines", "maximum_observed": 0, "passed": true}, {"first_repeated_ngram": null, "id": "repeated_20_token_ngram", "passed": true}, {"first_repeated_paragraph": null, "id": "repeated_paragraph", "passed": true}, {"finish_reason": "length", "id": "premature_eos", "passed": true, "task_completion_valid": false}, {"finish_reason": "length", "id": "runaway_generation", "output_token_cap": 4096, "output_tokens": 4096, "passed": false, "task_completion_valid": false}], "passed": false, "scorer": "corruption_and_repetition_v1"}, "decision_included": true, "diagnostic_only": false, "performance": {"decode_tokens_per_second": 32.0987050225712, "finish_reason": "length", "output_token_cap": 4096, "output_tokens": 4096, "prompt_tokens_per_second": 581.3528870451789, "request_count": 1, "telemetry": {"peak_process_rss_bytes": 6853292032, "peak_system_memory_used_bytes": 66999607296, "per_gpu": {"0": {"peak_memory_used_mib": 19831.0, "peak_power_draw_w": 166.03, "power_limit_w": 300.0, "pstates": ["P0"]}, "1": {"peak_memory_used_mib": 18265.0, "peak_power_draw_w": 155.6, "power_limit_w": 300.0, "pstates": ["P0"]}, "2": {"peak_memory_used_mib": 17851.0, "peak_power_draw_w": 149.68, "power_limit_w": 300.0, "pstates": ["P0"]}, "3": {"peak_memory_used_mib": 22029.0, "peak_power_draw_w": 152.5, "power_limit_w": 300.0, "pstates": ["P0"]}, "4": {"peak_memory_used_mib": 19417.0, "peak_power_draw_w": 151.84, "power_limit_w": 300.0, "pstates": ["P0"]}, "5": {"peak_memory_used_mib": 16847.0, "peak_power_draw_w": 156.64, "power_limit_w": 300.0, "pstates": ["P0"]}}, "sample_count": 117, "swap_delta_bytes": 0, "swap_used_end_bytes": 0, "swap_used_start_bytes": 0}, "time_to_first_token_seconds": 1.4199769360711798}, "quant": "UD-IQ3_XXS", "score": {"error": "no unified-diff file headers found", "max_score": 1.0, "passed": false, "score": 0.0, "scorer": "typescript_patch"}, "seed": 731, "status": "completed", "task": "typescript_patch", "visible_model_outputs": [{"bytes": 0, "path": "results/quants/UD-IQ3_XXS/cases/TS-PATCH-01/seed-731/attempts/attempt-001/turn-01/chat/model_output.bin", "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "utf8": ""}]}

Q8 LOAD-ONLY SMOKE
record: work/runpod-live/full-run-jsqxbdwx8mi8ti/pulled/results/quants/UD-Q8_K_XL/smoke-complete.json
record_sha256: c794a04a02e281fee7d301e9e1068a32dd58af53fa6c2a8551efd3da4ec0d7a8
{"completed_utc": "2026-08-01T23:37:37.787153+00:00", "excluded_from_quality_scoring": true, "health_ready": true, "label": "UD-Q8_K_XL", "last_health": "{\"status\":\"ok\"}", "load_seconds": 39.80830059701111, "passed": true, "phase": "smoke", "safe_to_release_model": true, "schema_version": 1, "server_command": {"argv": ["/root/kingy-llama.cpp/build/bin/llama-server", "--model", "/workspace/models/UD-Q8_K_XL/DeepSeek-V4-Flash-0731-UD-Q8_K_XL-00001-of-00005.gguf", "--jinja", "--ctx-size", "32768", "--parallel", "1", "--split-mode", "layer", "--n-gpu-layers", "all", "--flash-attn", "on", "--fit", "off", "--cache-type-k", "f16", "--cache-type-v", "f16", "--batch-size", "2048", "--ubatch-size", "2048", "--threads", "52", "--no-context-shift", "--no-warmup", "--metrics", "--temp", "1.0", "--top-p", "1.0", "--min-p", "0.0", "--top-k", "0", "--seed", "731", "--chat-template-kwargs", "{\"reasoning_effort\":\"high\"}", "--host", "127.0.0.1", "--port", "18080"], "context": 32768, "model": "/workspace/models/UD-Q8_K_XL/DeepSeek-V4-Flash-0731-UD-Q8_K_XL-00001-of-00005.gguf", "quant": "UD-Q8_K_XL", "runtime_commit": "c629da565c80b0b17fac6262acdca4d772e745d8", "shell": "/root/kingy-llama.cpp/build/bin/llama-server --model /workspace/models/UD-Q8_K_XL/DeepSeek-V4-Flash-0731-UD-Q8_K_XL-00001-of-00005.gguf --jinja --ctx-size 32768 --parallel 1 --split-mode layer --n-gpu-layers all --flash-attn on --fit off --cache-type-k f16 --cache-type-v f16 --batch-size 2048 --ubatch-size 2048 --threads 52 --no-context-shift --no-warmup --metrics --temp 1.0 --top-p 1.0 --min-p 0.0 --top-k 0 --seed 731 --chat-template-kwargs '{\"reasoning_effort\":\"high\"}' --host 127.0.0.1 --port 18080"}, "server_exit_code_after_controlled_stop": 0, "started_utc": "2026-08-01T23:36:57.978801+00:00", "status": "complete"}

EXPLICITLY NOT RUN OR NOT ESTABLISHED
- Quality scoring for UD-IQ1_S, UD-Q4_K_XL, or UD-Q8_K_XL
- A complete five-quant ranking or universal best-quant claim
- Any long-context retrieval, including 32K retrieval
- Maximum stable context or the 128K/384K/1,048,576-token ladder
- Controlled cold/warm load comparisons
- Native-model or official-endpoint reference
- The other eight repository labels
- Empirical validation on 96GB or 128GB systems
- Multilingual factual preservation, degradation, or general translation fluency