{ "complete": true, "evaluation_complete": true, "requested_tasks": [ "agi_eval_lsat_ar", "arc_challenge", "arc_easy", "bigbench_cs_algorithms", "bigbench_dyck_languages", "bigbench_language_identification", "bigbench_operators", "bigbench_qa_wikidata", "bigbench_repeat_copy_logic", "boolq", "commonsense_qa", "copa", "coqa", "hellaswag", "hellaswag_zeroshot", "jeopardy", "lambada_openai", "openbook_qa", "piqa", "squad", "winograd", "winogrande" ], "claim_status": "provisional_pending_prompt_review", "scores": { "hellaswag_zeroshot": 0.2410874366760254, "jeopardy": 0.0, "bigbench_qa_wikidata": 0.0, "arc_easy": 0.27146464586257935, "arc_challenge": 0.20989760756492615, "copa": 0.4699999988079071, "commonsense_qa": 0.1957411915063858, "piqa": 0.5174102187156677, "openbook_qa": 0.23399999737739563, "lambada_openai": 0.0, "hellaswag": 0.24158534407615662, "winograd": 0.5274725556373596, "winogrande": 0.519336998462677, "bigbench_dyck_languages": 0.0, "agi_eval_lsat_ar": 0.2347826063632965, "bigbench_cs_algorithms": 0.0060606058686971664, "bigbench_operators": 0.02857142873108387, "bigbench_repeat_copy_logic": 0.0, "squad": 0.0, "coqa": 0.014155079610645771, "boolq": 0.3782874643802643, "bigbench_language_identification": 0.2500999867916107 }, "aggregate": { "Core_v2": -0.04343042332051612, "centered_tasks": { "hellaswag_zeroshot": -0.011883417765299479, "jeopardy": 0.0, "bigbench_qa_wikidata": 0.0, "arc_easy": 0.02861952781677246, "arc_challenge": -0.05346985658009847, "copa": -0.06000000238418579, "commonsense_qa": -0.3471671834063889, "piqa": 0.03482043743133545, "openbook_qa": -0.02133333683013916, "lambada_openai": 0.0, "hellaswag": -0.011219541231791178, "winograd": 0.05494511127471924, "winogrande": 0.038673996925354004, "bigbench_dyck_languages": 0.0, "agi_eval_lsat_ar": -0.02028985818227132, "bigbench_cs_algorithms": 0.0060606058686971664, "bigbench_operators": 0.02857142873108387, "bigbench_repeat_copy_logic": 0.0, "squad": 0.0, "coqa": 0.014155079610645771, "boolq": -0.636085620051936, "bigbench_language_identification": 0.0001333157221476237 }, "protocol_id": "dclm-core-v2-361714bd-context2048-proposed", "required_tasks": 22 }, "lineage": { "counters": { "updates": 191, "tokens": 50000000 }, "manifests": { "recipe": "3fea67860037c274f5f1e5a5cb48349b4a5ef9345466084e8400bc398547826f", "data": "0c6c0d7bba82974895e1439a84e3f7515d0fadb54d2a18c9d05beb49efe0d641", "tokenizer": "d316298eef9ad7963ac948139d2284b35fc4e8c0a33776062f101a22b296a4fe", "code": "39ef30aa46adb8f8284e49097e0b7ae253aee5d7c4a2211e03b8abb1f4431ab6", "torch": "2.11.0+cu130" }, "schedule": { "warmup_tokens": 5000000, "peak_lr": 0.001, "fork_tokens": null, "end_tokens": null, "floor_ratio": 0.05 }, "checkpoint_sha256": "5f8daa094efc741e6f3c73ae6a91a8a76c8630fe6cd610e86c737fe8cf37d24e", "branch": null }, "prompt_audit_sha256": "6a895f22f85246bc077ae1ba0ebb13d35b8ae0c69002be136ffd01e538e5fef9", "item_ledger_sha256": "cebc0106e9725cfa95d42184543ab2dae6419406bf8b2340d7012584d7e2fcf6", "item_identity": "task, source file SHA256, zero-based source row, raw source row SHA256", "protocol_sha256": "f420fd3c2ab47c852633f60824e1f33a77f24a49baa841c0ff20610238bba345", "data_manifest_sha256": "0a750ef0ccb2e118cd467251faf07b9dc3d774662b9c1fdea807d9251c34c7a5", "tokenizer_policy": "BOS postprocessor; native Foundry context/answer tokenization", "precision": "fp32", "tf32": false, "elapsed_seconds": 22724.797448242083, "peak_allocated_bytes": 3860806656, "launch_manifest": { "scope": "experimental full E01 evaluation with per-example outcomes; 50M training targets and incomplete training decontamination", "source_snapshot": "/data/rfc005/full-core-e01-v1/src; byte-identical E01 source snapshot", "source_sha256": { "fabryka_track.english_base.icl": "c496f571bd78118f306479c62528c67d025c4ff1eca5594198388555230ee1b0", "fabryka_track.english_base.reference": "a8f75bd0134858c4c4cb16d789bfba69316afe480d29960c55fa9fd3ecfac2e6", "fabryka_track.english_base.prompt_audit": "7a891094a73bb9268c4d4bd2ce349d921917196447b57deae11ce2e351fc2af5", "fabryka_track.english_base.scores": "fa4226f90415195f008b7188b3df535aded2dc4039d0b44335ca7a35aa47fd4d", "fabryka_track.english_base.item_ledger": "352120820d0fd6333ad4897f966a57dda17e2b40742bdcf13aab9c7519e6e675" }, "protocol_sha256": "f420fd3c2ab47c852633f60824e1f33a77f24a49baa841c0ff20610238bba345", "data_manifest_sha256": "0a750ef0ccb2e118cd467251faf07b9dc3d774662b9c1fdea807d9251c34c7a5", "model_bundle_sha256": "ee046f48c84be5cad1af96958fc26792f6ebc2aad99605da13e4aedd87eebd90", "harness_sha256": "74aebbcd97945ca64688c56e1e51cd5a6d595f3f791441df2afa00227b3be41a", "precision": "fp32; TF32 disabled", "gpu": "NVIDIA GeForce RTX 3090", "torch": "2.3.1+cu121" }, "prompt_audit": { "hellaswag_zeroshot": { "trim_calls": 40168, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "jeopardy": { "trim_calls": 2117, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "bigbench_qa_wikidata": { "trim_calls": 20321, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "arc_easy": { "trim_calls": 9504, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "arc_challenge": { "trim_calls": 4688, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "copa": { "trim_calls": 200, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "commonsense_qa": { "trim_calls": 4884, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "piqa": { "trim_calls": 3676, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "openbook_qa": { "trim_calls": 2000, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "lambada_openai": { "trim_calls": 5153, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "hellaswag": { "trim_calls": 40168, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "winograd": { "trim_calls": 546, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "winogrande": { "trim_calls": 2534, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "bigbench_dyck_languages": { "trim_calls": 1000, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "agi_eval_lsat_ar": { "trim_calls": 920, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "bigbench_cs_algorithms": { "trim_calls": 1320, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "bigbench_operators": { "trim_calls": 210, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "bigbench_repeat_copy_logic": { "trim_calls": 32, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "squad": { "trim_calls": 10570, "truncated_calls": 6902, "lost_demonstrations_per_continuation": { "fully_removed": 5139, "partially_removed": 6864, "affected_continuations": 6889 }, "status": "measured; requires protocol review" }, "coqa": { "trim_calls": 7983, "truncated_calls": 0, "lost_demonstrations_per_continuation": { "fully_removed": 0, "partially_removed": 0, "affected_continuations": 0 }, "status": "measured; requires protocol review" }, "boolq": { "trim_calls": 6540, "truncated_calls": 338, "lost_demonstrations_per_continuation": { "fully_removed": 242, "partially_removed": 332, "affected_continuations": 334 }, "status": "measured; requires protocol review" }, "bigbench_language_identification": { "trim_calls": 40000, "truncated_calls": 7360, "lost_demonstrations_per_continuation": { "fully_removed": 12308, "partially_removed": 7288, "affected_continuations": 7332 }, "status": "measured; requires protocol review" } }, "artifact": { "local_path": "/root/data/rfc005/full-core-e01-v1/full-core.json", "sha256": "090942c43a7cf5f3af376ac54f6bcd78bc99a1155fbe4c01d2935a236014b2d1", "bytes": 6825705 }, "verification": { "all_source_continuations_accounted_for": true, "prompt_records": 204534, "truncated_continuations": 14600, "CORE_recomputed": true, "scope": "artifact integrity, original source identities, complete native item coverage, category-weighted scores, prompt audit and CORE arithmetic; experimental reference, not candidate quality or contamination proof", "original_item_identities_verified": true, "item_scores_recomputed": true, "item_records": 91037 } }