{ "arm": { "cards": [ "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a", "GPU-7aa0be10-974f-6430-52fe-09018a0e2e08" ], "host": "http://127.0.0.1:8000", "mode": "generate", "model": "gemma4-fp8-vllm", "name": "gemma4-fp8-generate", "note": "ARM 10: the base model under vLLM on the same two 3090s, so the gap to OpenJev is read without the runtime in it", "num_ctx": 16384, "quant": "FP8 dynamic (gemma-4-26B-A4B-it-FP8-dynamic, the estate's seat checkpoint)", "runtime": "vllm", "seed": 20260921, "top_logprobs": 20 }, "census": null, "ended": "2026-09-21T22:01:49Z", "energy": { "cards": [ "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a", "GPU-7aa0be10-974f-6430-52fe-09018a0e2e08" ], "decisions": 108, "joules": 2990.798844563961, "joules_per_decision": 27.692581894110752, "mean_watts": 426.28263004211476, "net_joules_per_decision": 11.00785136670199, "peak_watts": 240.39, "per_card": { "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a": { "joules": 1451.3871412968635, "mean_watts": 206.86818470786721, "min_watts": 106.67, "peak_watts": 228.89, "samples": 8, "seconds": 7.015999794006348 }, "GPU-7aa0be10-974f-6430-52fe-09018a0e2e08": { "joules": 1539.4117032670974, "mean_watts": 219.66491522429888, "min_watts": 128.47, "peak_watts": 240.39, "samples": 8, "seconds": 7.007999897003174 } }, "samples": 16, "seconds": 7.015999794006348 }, "idle": { "cards": [ "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a", "GPU-7aa0be10-974f-6430-52fe-09018a0e2e08" ], "joules": 2314.3358453643323, "mean_watts": 256.8345139490345, "peak_watts": 196.79, "per_card": { "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a": { "joules": 1064.5823661649226, "mean_watts": 118.14253109389912, "min_watts": 106.56, "peak_watts": 179.16, "samples": 10, "seconds": 9.011000156402588 }, "GPU-7aa0be10-974f-6430-52fe-09018a0e2e08": { "joules": 1249.7534791994094, "mean_watts": 138.76898438556864, "min_watts": 128.18, "peak_watts": 196.79, "samples": 10, "seconds": 9.00600004196167 } }, "samples": 20, "seconds": 9.011000156402588, "what": "the draw of these cards immediately before the arm, with the model resident and no work in flight" }, "kit": { "a_labels": { "no": 21, "yes": 42 }, "b_labelled": 1, "bank_sha256": "35d58866da02e3ac731814e63ef0ceac97a25ee8b224c1c32c29df6ddbcc3d1c", "built": null, "c_per_guest": 18, "c_roster": [ { "id": "darwin", "name": "Charles Darwin" }, { "id": "einstein", "name": "Albert Einstein" }, { "id": "hypatia", "name": "Hypatia of Alexandria" }, { "id": "ibn_sina", "name": "Ibn Sina" }, { "id": "sagan", "name": "Carl Sagan" }, { "id": "socrates", "name": "Socrates of Athens" } ], "counts": { "a": 63, "articles": 42, "b": 42, "c": 108 }, "docroot": "/workshop/s2s-research-hub/site", "seed": 20260921, "wall_sha256": "80484f8379a5977cf56033f335c27c8f54fac683c7adf3e800def5fa419e698c", "wall_url": "https://longtable.strata2signal.com/api/wall" }, "overlong": {}, "pin": { "card": "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a", "cards": [ "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a", "GPU-7aa0be10-974f-6430-52fe-09018a0e2e08" ], "floor_mib": 200.0, "growth_mib": { "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a": 0.0, "GPU-7aa0be10-974f-6430-52fe-09018a0e2e08": 0.0 }, "memory_after_mib": { "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a": 22050.0, "GPU-7aa0be10-974f-6430-52fe-09018a0e2e08": 22050.0 }, "memory_after_named_mib": 22050.0, "memory_after_per_card_mib": { "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a": 22050.0, "GPU-7aa0be10-974f-6430-52fe-09018a0e2e08": 22050.0 }, "named_grew": false, "ok": false, "others_quiet": true, "per_card_grew": { "GPU-46890836-d8f7-e868-1977-c1ce28db0d7a": false, "GPU-7aa0be10-974f-6430-52fe-09018a0e2e08": false }, "resident": true }, "readout": { "NOUL_BIAS": 0.0, "NOUL_T": 1.829074, "T": 0.85, "floor": -30.0, "instructions": { "a": "Does the article in the state answer this reader's question? Question: %s", "b": "Which section of the article in the state contains the answer to this reader's question? Question: %s", "c": "Six people are at a dinner table. Which of them said the line below?\nLine: %s" }, "protocol": "openjev old (untargeted) path" }, "refusal_policy": "README §A2.5: an item whose prompt exceeds the served context is refused with its measured token count and excluded from every rate; it is never truncated", "started": "2026-09-21T22:01:31Z", "summary": { "accuracy": 0.8240740740740741, "brier": 0.3487654320987654, "correct": 89, "errors": 0, "eval_tokens_total": 442, "floored_rows": 0, "letters_floored_total": 0, "median_seconds": 0.06399377600064327, "n": 108, "n_labelled": 108, "n_reached": 108, "p95_seconds": 0.0863841649988899, "prompt_tokens_total": 30905, "refused": 0, "refused_by_reason": {}, "reliability": [ { "accuracy": null, "correct": 0, "gap": null, "hi": 0.1, "lo": 0.0, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": 0.5, "correct": 1, "gap": -0.33333333333333337, "hi": 0.2, "lo": 0.1, "mean_p": 0.16666666666666666, "n": 2, "sum_p": 0.3333333333333333 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.3, "lo": 0.2, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.4, "lo": 0.3, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.5, "lo": 0.4, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.6, "lo": 0.5, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.7, "lo": 0.6, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.8, "lo": 0.7, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": null, "correct": 0, "gap": null, "hi": 0.9, "lo": 0.8, "mean_p": null, "n": 0, "sum_p": 0.0 }, { "accuracy": 0.8301886792452831, "correct": 88, "gap": 0.16981132075471694, "hi": 1.0, "lo": 0.9, "mean_p": 1.0, "n": 106, "sum_p": 106.0 } ], "unparsed": 2 }, "task": "c", "wall_seconds": 7.239073123997514 }