{
  "schema_version": 1,
  "adapter_key": "soar",
  "validated_at": "2026-10-14",
  "seeds": [
    42,
    43,
    44
  ],
  "accelerator": {
    "vendor": "NVIDIA",
    "model": "NVIDIA-A100-SXM4-80GB"
  },
  "command": [
    "python",
    "scripts/run_oct13_post_training.py",
    "--method",
    "soar",
    "--config",
    "data/oct14_execution-soar7b-full-v1.json",
    "--seed",
    "42",
    "--output",
    "runs/soar-source-consistent-s42",
    "--variant",
    "source-consistent"
  ],
  "all_sanitized_commands": [
    [
      "python",
      "scripts/run_oct13_post_training.py",
      "--method",
      "soar",
      "--config",
      "data/oct14_execution-soar7b-full-v1.json",
      "--seed",
      "42",
      "--output",
      "runs/soar-source-consistent-s42",
      "--variant",
      "source-consistent"
    ],
    [
      "python",
      "scripts/run_oct13_post_training.py",
      "--method",
      "soar",
      "--config",
      "data/oct14_execution-soar7b-full-v1.json",
      "--seed",
      "42",
      "--output",
      "runs/soar-without-source-reward-s42",
      "--variant",
      "without-source-reward"
    ],
    [
      "python",
      "scripts/run_oct13_post_training.py",
      "--method",
      "soar",
      "--config",
      "data/oct14_execution-soar7b-full-v1.json",
      "--seed",
      "43",
      "--output",
      "runs/soar-source-consistent-s43",
      "--variant",
      "source-consistent"
    ],
    [
      "python",
      "scripts/run_oct13_post_training.py",
      "--method",
      "soar",
      "--config",
      "data/oct14_execution-soar7b-full-v1.json",
      "--seed",
      "43",
      "--output",
      "runs/soar-without-source-reward-s43",
      "--variant",
      "without-source-reward"
    ],
    [
      "python",
      "scripts/run_oct13_post_training.py",
      "--method",
      "soar",
      "--config",
      "data/oct14_execution-soar7b-full-v1.json",
      "--seed",
      "44",
      "--output",
      "runs/soar-source-consistent-s44",
      "--variant",
      "source-consistent"
    ],
    [
      "python",
      "scripts/run_oct13_post_training.py",
      "--method",
      "soar",
      "--config",
      "data/oct14_execution-soar7b-full-v1.json",
      "--seed",
      "44",
      "--output",
      "runs/soar-without-source-reward-s44",
      "--variant",
      "without-source-reward"
    ]
  ],
  "command_path_relocations_only": true,
  "dataset": {
    "verified": true,
    "test_read": false,
    "source_id": "Elliotes/EcoAgent#derived-tool-document-attribute-training-QA-v1",
    "source_revision": "bc5db40a20b254c5249401d2220ea71b06c86a5c",
    "source_sha256": "24b94a6905137c6835c4ff9afbf418d833d0d854505686295caee8f7608a7771",
    "partition_sha256": {
      "train": "973266e4083994c6eee1e84dabb748af028b8cd5e4f00e6201496152d7be93d5",
      "validation": "2d1c706f6e0d4a05eaaddb3ecbb16d7e325d35dab265612c308e99a011fb4f9e"
    },
    "partition": "tool UID sha256(soar-tool-group-v1|UID) unsigned 64-bit prefix <0.2 validation; original-order bounded prefixes"
  },
  "checkpoint": {
    "policy": {
      "public_id": "Qwen/Qwen2.5-7B-Instruct",
      "revision": "bb46c15ee4bb56c5b63245ef50fd7637234d6f75",
      "verified_mirror": {
        "public_id": "Qwen/Qwen2.5-7B-Instruct",
        "revision": "bb46c15ee4bb56c5b63245ef50fd7637234d6f75",
        "file_sha256": {
          "config.json": "7463bb0ea78315365e6c6b74de4e73bbcc8359dfb0c5a737584e077d42c0b03c",
          "generation_config.json": "3a8f9087e486054c8a4a08dae2e5a3ba62e23da212b5b8c08bc42cb983c3459f",
          "merges.txt": "599bab54075088774b1733fde865d5bd747cbcc7a547c5bc12610e874e26f5e3",
          "model-00001-of-00004.safetensors": "a1333e6293854747c481288ea83b348226af178dd565c49b6f9495ba1966aba7",
          "model-00002-of-00004.safetensors": "f5d25a2772cb825164a2a2c0fb6d51a87e282abf21e4dd75bc5cfb3cd0ea6185",
          "model-00003-of-00004.safetensors": "8efdec4c1bc12317ae1a38dc42b595ce777738a64deea3fcb8a0a91381bcdfd5",
          "model-00004-of-00004.safetensors": "1a72d403cdf0c1ec3cb7f289f17b394a01e64394c2e9b3c0f94dbce3faf879bd",
          "model.safetensors.index.json": "624bf7c47cd12468fdc16e38a47cf4f19e0415b859a223ba3c027eed2f0e1028",
          "tokenizer.json": "c0382117ea329cdf097041132f6d735924b697924d6f6fc3945713e96ce87539",
          "tokenizer_config.json": "5b5d4f65d0acd3b2d56a35b56d374a36cbc1c8fa5cf3b3febbbfabf22f359583",
          "vocab.json": "ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910"
        }
      }
    },
    "frozen_reference": {
      "public_id": "Qwen/Qwen2.5-7B-Instruct",
      "revision": "bb46c15ee4bb56c5b63245ef50fd7637234d6f75",
      "verified_mirror": {
        "public_id": "Qwen/Qwen2.5-7B-Instruct",
        "revision": "bb46c15ee4bb56c5b63245ef50fd7637234d6f75",
        "file_sha256": {
          "config.json": "7463bb0ea78315365e6c6b74de4e73bbcc8359dfb0c5a737584e077d42c0b03c",
          "generation_config.json": "3a8f9087e486054c8a4a08dae2e5a3ba62e23da212b5b8c08bc42cb983c3459f",
          "merges.txt": "599bab54075088774b1733fde865d5bd747cbcc7a547c5bc12610e874e26f5e3",
          "model-00001-of-00004.safetensors": "a1333e6293854747c481288ea83b348226af178dd565c49b6f9495ba1966aba7",
          "model-00002-of-00004.safetensors": "f5d25a2772cb825164a2a2c0fb6d51a87e282abf21e4dd75bc5cfb3cd0ea6185",
          "model-00003-of-00004.safetensors": "8efdec4c1bc12317ae1a38dc42b595ce777738a64deea3fcb8a0a91381bcdfd5",
          "model-00004-of-00004.safetensors": "1a72d403cdf0c1ec3cb7f289f17b394a01e64394c2e9b3c0f94dbce3faf879bd",
          "model.safetensors.index.json": "624bf7c47cd12468fdc16e38a47cf4f19e0415b859a223ba3c027eed2f0e1028",
          "tokenizer.json": "c0382117ea329cdf097041132f6d735924b697924d6f6fc3945713e96ce87539",
          "tokenizer_config.json": "5b5d4f65d0acd3b2d56a35b56d374a36cbc1c8fa5cf3b3febbbfabf22f359583",
          "vocab.json": "ca10d7e9fb3ed18575dd1e277a2579c16d108e32f27439684afa0e10b1440910"
        }
      }
    }
  },
  "result": "passed",
  "metrics": {
    "source-consistent.initial.accuracy": {
      "mean": 0.08333333333333333,
      "std": 0.14433756729740643,
      "n": 3
    },
    "source-consistent.initial.source_consistency": {
      "mean": 0.25,
      "std": 0,
      "n": 3
    },
    "source-consistent.initial.format_compliance": {
      "mean": 0.5833333333333334,
      "std": 0.14433756729740643,
      "n": 3
    },
    "source-consistent.initial.mean_searches": {
      "mean": 0.3333333333333333,
      "std": 0.14433756729740643,
      "n": 3
    },
    "source-consistent.validation.accuracy": {
      "mean": 0.3333333333333333,
      "std": 0.14433756729740643,
      "n": 3
    },
    "source-consistent.validation.source_consistency": {
      "mean": 0.4166666666666667,
      "std": 0.14433756729740643,
      "n": 3
    },
    "source-consistent.validation.format_compliance": {
      "mean": 0.75,
      "std": 0.25,
      "n": 3
    },
    "source-consistent.validation.mean_searches": {
      "mean": 0.6666666666666666,
      "std": 0.14433756729740643,
      "n": 3
    },
    "without-source-reward.initial.accuracy": {
      "mean": 0.08333333333333333,
      "std": 0.14433756729740643,
      "n": 3
    },
    "without-source-reward.initial.source_consistency": {
      "mean": 0.25,
      "std": 0,
      "n": 3
    },
    "without-source-reward.initial.format_compliance": {
      "mean": 0.5833333333333334,
      "std": 0.14433756729740643,
      "n": 3
    },
    "without-source-reward.initial.mean_searches": {
      "mean": 0.3333333333333333,
      "std": 0.14433756729740643,
      "n": 3
    },
    "without-source-reward.validation.accuracy": {
      "mean": 0.16666666666666666,
      "std": 0.14433756729740643,
      "n": 3
    },
    "without-source-reward.validation.source_consistency": {
      "mean": 0.25,
      "std": 0,
      "n": 3
    },
    "without-source-reward.validation.format_compliance": {
      "mean": 0.5833333333333334,
      "std": 0.14433756729740643,
      "n": 3
    },
    "without-source-reward.validation.mean_searches": {
      "mean": 0.4166666666666667,
      "std": 0.28867513459481287,
      "n": 3
    }
  },
  "execution_checks": {
    "per_seed": {
      "42": {
        "source-consistent": {
          "actual_optimizer_updates": 4,
          "equal_reward_groups_skipped": 0,
          "all_four_groups_audited": true,
          "reward_derivation_audits": [
            {
              "step": 1,
              "derived_rewards": [
                0.23076923076923078,
                0.23076923076923078,
                0.15384615384615385,
                0.23076923076923078
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 2,
              "derived_rewards": [
                0.23076923076923078,
                0,
                0,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 3,
              "derived_rewards": [
                0.23076923076923078,
                0.23076923076923078,
                0,
                0.23076923076923078
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 4,
              "derived_rewards": [
                0,
                0,
                0.15384615384615385,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            }
          ],
          "invocation_elapsed_seconds": 87.91338889487088,
          "configured_wall_cap_seconds": 1800,
          "measured_cost": {
            "generation_calls": {
              "policy_trajectory": 35,
              "isolated_internal_recall": 11
            },
            "generated_tokens": {
              "policy_trajectory": 954,
              "isolated_internal_recall": 408
            },
            "scored_response_tokens": {
              "behavior": 632,
              "student_rl": 632,
              "reference": 632
            },
            "frozen_encoder_forward_calls": 0,
            "frozen_encoder_input_tokens_including_prompt": 0,
            "scope": "entire invocation, including training controls, profiling and evaluation",
            "includes_rejected_and_profiling_generations": true
          }
        },
        "without-source-reward": {
          "actual_optimizer_updates": 3,
          "equal_reward_groups_skipped": 1,
          "all_four_groups_audited": true,
          "reward_derivation_audits": [
            {
              "step": 1,
              "derived_rewards": [
                0.16666666666666669,
                0.16666666666666669,
                0.16666666666666669,
                0.16666666666666669
              ],
              "mathematically_equal": true,
              "optimizer_update_executed": false
            },
            {
              "step": 2,
              "derived_rewards": [
                0.16666666666666669,
                0,
                0.16666666666666669,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 3,
              "derived_rewards": [
                0.16666666666666669,
                0,
                0,
                0.16666666666666669
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 4,
              "derived_rewards": [
                0,
                0.16666666666666669,
                0,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            }
          ],
          "invocation_elapsed_seconds": 85.92835174314678,
          "configured_wall_cap_seconds": 1800,
          "measured_cost": {
            "generation_calls": {
              "policy_trajectory": 34,
              "isolated_internal_recall": 10
            },
            "generated_tokens": {
              "policy_trajectory": 1002,
              "isolated_internal_recall": 364
            },
            "scored_response_tokens": {
              "behavior": 705,
              "student_rl": 513,
              "reference": 513
            },
            "frozen_encoder_forward_calls": 0,
            "frozen_encoder_input_tokens_including_prompt": 0,
            "scope": "entire invocation, including training controls, profiling and evaluation",
            "includes_rejected_and_profiling_generations": true
          }
        }
      },
      "43": {
        "source-consistent": {
          "actual_optimizer_updates": 4,
          "equal_reward_groups_skipped": 0,
          "all_four_groups_audited": true,
          "reward_derivation_audits": [
            {
              "step": 1,
              "derived_rewards": [
                0.15384615384615385,
                0.23076923076923078,
                0,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 2,
              "derived_rewards": [
                0,
                0,
                0,
                0.15384615384615385
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 3,
              "derived_rewards": [
                0.15384615384615385,
                0,
                0.23076923076923078,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 4,
              "derived_rewards": [
                0.23076923076923078,
                0.15384615384615385,
                0.15384615384615385,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            }
          ],
          "invocation_elapsed_seconds": 87.62172612734139,
          "configured_wall_cap_seconds": 1800,
          "measured_cost": {
            "generation_calls": {
              "policy_trajectory": 36,
              "isolated_internal_recall": 12
            },
            "generated_tokens": {
              "policy_trajectory": 1004,
              "isolated_internal_recall": 382
            },
            "scored_response_tokens": {
              "behavior": 707,
              "student_rl": 707,
              "reference": 707
            },
            "frozen_encoder_forward_calls": 0,
            "frozen_encoder_input_tokens_including_prompt": 0,
            "scope": "entire invocation, including training controls, profiling and evaluation",
            "includes_rejected_and_profiling_generations": true
          }
        },
        "without-source-reward": {
          "actual_optimizer_updates": 4,
          "equal_reward_groups_skipped": 0,
          "all_four_groups_audited": true,
          "reward_derivation_audits": [
            {
              "step": 1,
              "derived_rewards": [
                0.16666666666666669,
                0.16666666666666669,
                0,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 2,
              "derived_rewards": [
                0,
                0.16666666666666669,
                0,
                0.16666666666666669
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 3,
              "derived_rewards": [
                0,
                0,
                0.16666666666666669,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 4,
              "derived_rewards": [
                0.16666666666666669,
                0.16666666666666669,
                0.16666666666666669,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            }
          ],
          "invocation_elapsed_seconds": 86.7255923692137,
          "configured_wall_cap_seconds": 1800,
          "measured_cost": {
            "generation_calls": {
              "policy_trajectory": 36,
              "isolated_internal_recall": 12
            },
            "generated_tokens": {
              "policy_trajectory": 973,
              "isolated_internal_recall": 401
            },
            "scored_response_tokens": {
              "behavior": 676,
              "student_rl": 676,
              "reference": 676
            },
            "frozen_encoder_forward_calls": 0,
            "frozen_encoder_input_tokens_including_prompt": 0,
            "scope": "entire invocation, including training controls, profiling and evaluation",
            "includes_rejected_and_profiling_generations": true
          }
        }
      },
      "44": {
        "source-consistent": {
          "actual_optimizer_updates": 4,
          "equal_reward_groups_skipped": 0,
          "all_four_groups_audited": true,
          "reward_derivation_audits": [
            {
              "step": 1,
              "derived_rewards": [
                0,
                0,
                0,
                0.23076923076923078
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 2,
              "derived_rewards": [
                0,
                0,
                0.15384615384615385,
                0.23076923076923078
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 3,
              "derived_rewards": [
                0,
                0.23076923076923078,
                0,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 4,
              "derived_rewards": [
                0.15384615384615385,
                0,
                0,
                0
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            }
          ],
          "invocation_elapsed_seconds": 77.97193231619895,
          "configured_wall_cap_seconds": 1800,
          "measured_cost": {
            "generation_calls": {
              "policy_trajectory": 32,
              "isolated_internal_recall": 8
            },
            "generated_tokens": {
              "policy_trajectory": 844,
              "isolated_internal_recall": 222
            },
            "scored_response_tokens": {
              "behavior": 519,
              "student_rl": 519,
              "reference": 519
            },
            "frozen_encoder_forward_calls": 0,
            "frozen_encoder_input_tokens_including_prompt": 0,
            "scope": "entire invocation, including training controls, profiling and evaluation",
            "includes_rejected_and_profiling_generations": true
          }
        },
        "without-source-reward": {
          "actual_optimizer_updates": 3,
          "equal_reward_groups_skipped": 1,
          "all_four_groups_audited": true,
          "reward_derivation_audits": [
            {
              "step": 1,
              "derived_rewards": [
                0,
                0,
                0,
                0.16666666666666669
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 2,
              "derived_rewards": [
                0,
                0,
                0.16666666666666669,
                0.16666666666666669
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 3,
              "derived_rewards": [
                0,
                0.16666666666666669,
                0,
                0.16666666666666669
              ],
              "mathematically_equal": false,
              "optimizer_update_executed": true
            },
            {
              "step": 4,
              "derived_rewards": [
                0,
                0,
                0,
                0
              ],
              "mathematically_equal": true,
              "optimizer_update_executed": false
            }
          ],
          "invocation_elapsed_seconds": 79.04650752060115,
          "configured_wall_cap_seconds": 1800,
          "measured_cost": {
            "generation_calls": {
              "policy_trajectory": 31,
              "isolated_internal_recall": 7
            },
            "generated_tokens": {
              "policy_trajectory": 932,
              "isolated_internal_recall": 208
            },
            "scored_response_tokens": {
              "behavior": 634,
              "student_rl": 412,
              "reference": 412
            },
            "frozen_encoder_forward_calls": 0,
            "frozen_encoder_input_tokens_including_prompt": 0,
            "scope": "entire invocation, including training controls, profiling and evaluation",
            "includes_rejected_and_profiling_generations": true
          }
        }
      }
    }
  },
  "executed_source_sha256": {
    "scripts/prepare_soar_catalog_qa.py": "483a4eb3d9932d31046888c165c06590e67c6df5575a5b2afd4e302145f445e7",
    "scripts/run_oct13_post_training.py": "916b3b9a0eaeeaf30bad811376fa05305011754c922bb368a0f39833aad07ffa",
    "src/auto_research/oct12_public_sources.py": "4e1150cbe4ea975819c244267180f853c83c4ad4c2c27f9e0ce95cc16793e093",
    "src/auto_research/post_training/oct10_checkpoint.py": "247f1f97a3ca701277f241ec81d99296f0ea82e1db8f68006fcd7bfdc661cd74",
    "src/auto_research/post_training/oct13_common.py": "d19401266a420afdc64867f145f8384fd649dc1991e4b74ea1f7ee8b8d91bde5",
    "src/auto_research/post_training/proximal_checkpoint.py": "5307a54204720e5407d2cd0435b654ae0e594f30c85cdf07d55d4e90612a1397",
    "src/auto_research/post_training/response_policy.py": "e380a5b4d1d53974b973b72b7d9ad6328129dc8c9bae4d233f6c1cbdf8508bf7",
    "src/auto_research/public_holdout.py": "033d9c1fd112accf5ac79573b99d9f31cffc646bcab6911cf3e49459fa003b46",
    "src/auto_research/reproductions/gppo_framework/__init__.py": "7fe8cab42fef9f32b9653dfb5f54f03182de0753de2c9e5b359d124d7cd763d8",
    "src/auto_research/reproductions/gppo_framework/adapter.py": "a7f4ee792a8ded3daeeb183cbca5c99de63c1b7d3cb195f5bedd5289cb0e0260",
    "src/auto_research/reproductions/gppo_framework/objective.py": "0f8fbb2ad397b04a70466dde4e939f4d094ad7e7df07cfbba59f4a6ca5d05dd0",
    "src/auto_research/reproductions/gppo_framework/paper.yaml": "a023ee98bbd078077b2708122973d27a98085848be816ce461b7d4291aca8e1f",
    "src/auto_research/reproductions/grpo_difficulty_curriculum/__init__.py": "d09d8d7e0e6f5bcec9979241eed06a1d2a6ac8f3fb7a25f49e0df55be0071acb",
    "src/auto_research/reproductions/grpo_difficulty_curriculum/adapter.py": "bbd3e964dc035ea4bbd999a581ea07eccabd986b6456774b21a4c24092cbdfbe",
    "src/auto_research/reproductions/grpo_difficulty_curriculum/algorithm.py": "4131e39279d30c0748b14f611ab657b3fea76147a7c164aea2883c62bde75785",
    "src/auto_research/reproductions/grpo_difficulty_curriculum/checkpoint.py": "d03a7d2db0a19b469fe5429ddce634b02d8cbd0eeed8d09bef0afc51c1d7ac70",
    "src/auto_research/reproductions/grpo_difficulty_curriculum/paper.yaml": "c417d2269314b15e2dfef4ba52dfb24771b2e1270f808fe5be47a54e9d0c1bb1",
    "src/auto_research/reproductions/magicbeanstalk/__init__.py": "8dd207ea37a89dfa34f87b02eb4a2fd26472895678d1362a7e5b59dad5cda616",
    "src/auto_research/reproductions/magicbeanstalk/adapter.py": "319b6503de342feee49fb6446b7bbdeb117f7dc0dd7b79ed1c3bd34cce1ce73f",
    "src/auto_research/reproductions/magicbeanstalk/algorithm.py": "2fb98f7137f336414ae6e07069fe34584d0e1cbc4bd594683a67e11ffc982aba",
    "src/auto_research/reproductions/magicbeanstalk/checkpoint.py": "78416567c61f23505f579805b845a40a3098c311e57fb42c7ac8d5ebc67390dd",
    "src/auto_research/reproductions/magicbeanstalk/paper.yaml": "e9c5619918b4d72d776008aba3fa95d50ba6ebe99afab729ba9e6dfe82fac06e",
    "src/auto_research/reproductions/progressive_contraction/__init__.py": "e50fe46d888e0a0b11b7d4d6d8f3224ae5940394f8ba8396fc5afb1cd38a4d53",
    "src/auto_research/reproductions/progressive_contraction/adapter.py": "06f23b5c795f7222f76dee9beaa8164f460eb5183e18c6d374bff011737235dd",
    "src/auto_research/reproductions/progressive_contraction/algorithm.py": "ba077b8c52a8d10e268a4e5c85770b9cebcb12415f9b2aaf7b9500e1303ca4b1",
    "src/auto_research/reproductions/progressive_contraction/checkpoint.py": "67fa44814d613d3838716fea68476217d608af14f23a051e32a916b72538e7ec",
    "src/auto_research/reproductions/progressive_contraction/paper.yaml": "faf5243c71a87407601b982469d98dc4d48ddf842da56c210ede077c2b27c1c1",
    "src/auto_research/reproductions/soar/__init__.py": "3d87d1189d7037e32c1e2cb0c249d13609c5af3d7062a19904590ba7923c799a",
    "src/auto_research/reproductions/soar/adapter.py": "8fbbe8f8fe54b9f014ccc26a75b01e72b73b95e2b03cc379c5b5db4b63c9b31e",
    "src/auto_research/reproductions/soar/algorithm.py": "9b7068d9774d1c93800c512a04b9ba4d0ee98fc13d308e5ac0970299c8bb2485",
    "src/auto_research/reproductions/soar/checkpoint.py": "91901d9eb9f81faa41dda170ff1f4d2283f56cd17c7df0572348356c7b7eb76e",
    "src/auto_research/reproductions/soar/paper.yaml": "2f756cab9f7fefd4613bb69971fd8252c2b9ad451fa5c5eff5d03fa063d91521",
    "src/auto_research/reproductions/state_opd/__init__.py": "8db07736f0336ed7f39502ae2953bdae76658942542cacbe4b6b2c1fd0abe094",
    "src/auto_research/reproductions/state_opd/adapter.py": "2f8bf4773e8f40c0eabe777fbd1e1aa6d372a7b05501404c3e3f7b2d54987524",
    "src/auto_research/reproductions/state_opd/algorithm.py": "dc7fe3eae619ff69ec51c47826f89471f00be564ea4e6730e12726a060ba7e2f",
    "src/auto_research/reproductions/state_opd/checkpoint.py": "95755e0604f563a5c96f92e9de9069befe3e1274d31c224a43ca20d71f756b09",
    "src/auto_research/reproductions/state_opd/environment.py": "0e9564f9ec540985eb010b1e70f46938acc1bd946041b518d6579253762080e1",
    "src/auto_research/reproductions/state_opd/paper.yaml": "86c75255f1c416611a181c26359e9f5d9865caaea5266ca7b13dd8a183b906cc",
    "src/auto_research/reproductions/synthself/__init__.py": "474bb75e10b95afe6faec1f3631beab2f17f9a43013d3d29e79b7d4b36e06031",
    "src/auto_research/reproductions/synthself/adapter.py": "db4c08e1ccc2acfc28d19fc72acfa39b0a0dce096bbc430bafe9b21dbb72cddd",
    "src/auto_research/reproductions/synthself/algorithm.py": "1e7ea1ff65d3ba0ece1af889022c19f7147f51d10fb8196ab8f9fa2e18604b95",
    "src/auto_research/reproductions/synthself/checkpoint.py": "dbcf9f222094c197e0392f82191a0ae6551c6573ee54a9eee8c7d37553472981",
    "src/auto_research/reproductions/synthself/paper.yaml": "d6d31d755561626c2b667f4fc2b990dee8360fa52f92193b5c6a6e93b1f6ca71",
    "src/auto_research/reproductions/tapo_trajectory/__init__.py": "a0dd8d4825b91dd076b71bdd3a3eccb0638ba4bf0240fe9da76d61a7290f69c6",
    "src/auto_research/reproductions/tapo_trajectory/adapter.py": "c01b75bf7853546caa2d517dd5f9c9faf739537b95207b8f1fd28adcdd5aafa7",
    "src/auto_research/reproductions/tapo_trajectory/objective.py": "49ab02ac758251acde90091a82b22576449d294a1f25cc3144ff1addd1132ae2",
    "src/auto_research/reproductions/tapo_trajectory/paper.yaml": "493e59ec67f96ee37ffecf6694dd5c4d8555849b8f756ccb63644da90bbb48d6",
    "src/auto_research/reproductions/upturn/__init__.py": "1e9793e9ea85e94144f7eac8fa141567d7f964438fa628788b668358154e196e",
    "src/auto_research/reproductions/upturn/adapter.py": "72e5da3cb70856944d0c9a25376f73d4a1d9f16213efd43e19b3acf2a584afc2",
    "src/auto_research/reproductions/upturn/algorithm.py": "421abd06cc492c39cf72e9b15c7f85c01e85fae0a7c26fb3053fbd19bc4897a4",
    "src/auto_research/reproductions/upturn/checkpoint.py": "fbcb7c830cf53d3130cfc06621f3178516aa72c7a939f4f765c3c7864b383787",
    "src/auto_research/reproductions/upturn/paper.yaml": "6eca098f6d945a30732f54cb04f2ee984b165a3c4f06cb65d49879d7edd28696",
    "scripts/diagnose_oct14_post_training.py": "2a15d720a4f84bd1e415a4f0653c589c9716d526d4842acd5df728fd3575ca5b",
    "scripts/diagnose_oct14_soar_format.py": "02b375e4359b8b8031246853a7e30e52b0cf7e0a94cf563299758260953f7ded"
  },
  "limitations": [
    "Three actual seeds42/43/44 and two arms complete this reduced-scale mechanism plan, not the multi-method batch.",
    "Three seeds remain L1 public tool-catalog diagnostics, not original RATED/seven-QA benchmarks or stable capability evidence.",
    "Independent public tool-catalog attribute QA is not the original RATED or seven-QA benchmark; tiny4validation is L1 only.",
    "Real LoRA policy, isolated same-policy internal recall, lexical retrieval and frozen-reference trajectory GRPO are executed.",
    "Raw zero groups and all actual generation/scoring costs are retained; matched configured budget is not identical realized compute.",
    "The measured source-match difference is not a stable capability or production improvement claim."
  ],
  "provenance": {
    "commit": "b90862754bf51b1beeb5add3f7dc2a51ca00ace4",
    "artifact_path": "docs/gpu-validations/soar-a100-20261014-seeds42-44.json",
    "metrics_artifact": "docs/reproductions/2610.13157-soar/metrics/public-a100-complete-seeds42-44-b9086275.json",
    "execution_config_sha256": "5ee7cfb10fe21078104f439a8de1bf48a528a10a138d08bd26ccbf030b1185a3",
    "raw_results_sha256": {
      "42.source-consistent": "f0597403b4e7f02ff5e9e5c70def793f5c4c8c38ea406efbcbd86e9bd8267d9b",
      "42.without-source-reward": "3e8f9a849a9e9aa2557b0ee0d15fc270dce73406b2d7a650c03980f08ccd5409",
      "43.source-consistent": "11d2406c6a0a60a8399144d6618d186f7f38be00a16d9ac2819b1cc46696feb9",
      "43.without-source-reward": "f9c7d3bc6915dfbd56beffb5fa21628fb38ffa14a197eb99f0c29f01367a7b5a",
      "44.source-consistent": "94124e6e6309f33fbdbe9a62e5681b3fd9cf0525b22c470259da1fa41099292b",
      "44.without-source-reward": "d84a170887a24364770bc518b01b2d7e78b663da38163588045e4cdf71b6d088"
    },
    "checkpoint_committed": false
  }
}