{
  "schema_version": 1,
  "adapter_key": "near",
  "validated_at": "2026-10-13T17:04:36.967501+00:00",
  "accelerator": {
    "vendor": "NVIDIA",
    "model": "NVIDIA A30"
  },
  "command": [
    "/tmp/oct14-memory-overlay-4496283a/bin/python",
    "scripts/run_oct13_memory.py",
    "--paper",
    "near",
    "--raw-data",
    "/tmp/auto-research-oct13-ccad3e67/public/longmemeval/longmemeval_s_cleaned.json",
    "--model-dir",
    "/tmp/auto-research-oct13-ccad3e67/public/minilm",
    "--device",
    "cuda",
    "--seed",
    "42",
    "--max-queries",
    "100",
    "--max-length",
    "128",
    "--batch-size",
    "16",
    "--adapter-epochs",
    "2",
    "--split-policy",
    "query-disjoint-shared-bank",
    "--diagnostic-only",
    "--output",
    "/tmp/auto-research-oct14-34148c2e/runs/memory-oct14/near-query-disjoint-shared-bank-100q-seed42.json"
  ],
  "seeds": [
    42,
    43,
    44
  ],
  "dataset": {
    "kind": "dataset",
    "public_id": "xiaowu0162/longmemeval-cleaned",
    "revision": "98d7416c24c778c2fee6e6f3006e7a073259d48f",
    "upstream_file": "longmemeval_s_cleaned.json",
    "transformation": "none; complete pinned public raw file",
    "files": {
      "longmemeval_s_cleaned.json": "d6f21ea9d60a0d56f34a05b609c79c88a451d2ae03597821ea3d5a9678c3a442"
    },
    "source_verification": {
      "immutable_revision_verified": true,
      "verification": "existing public-source pin and local byte-hash verification before transfer"
    }
  },
  "checkpoint": {
    "kind": "checkpoint",
    "public_id": "sentence-transformers/all-MiniLM-L6-v2",
    "revision": "1110a243fdf4706b3f48f1d95db1a4f5529b4d41",
    "files": {
      "model.safetensors": "53aa51172d142c89d9012cce15ae4d6cc0ca6895895114379cacb4fab128d9db",
      "tokenizer_config.json": "acb92769e8195aabd29b7b2137a9e6d6e25c476a4f15aa4355c233426c61576b",
      "special_tokens_map.json": "303df45a03609e4ead04bc3dc1536d0ab19b5358db685b6f3da123d05ec200e3",
      "config.json": "953f9c0d463486b10a6871cc2fd59f223b2c70184f49815e7efbcab5d8908b41",
      "tokenizer.json": "be50c3628f2bf5bb5e3a7f17b1f74611b2561a3a27eeab05e5aa30f411572037",
      "vocab.txt": "07eced375cec144d27c900241f3e339478dec958f92fddbc551f295c992038a3"
    },
    "source_verification": {
      "id": "sentence-transformers/all-MiniLM-L6-v2",
      "revision": "1110a243fdf4706b3f48f1d95db1a4f5529b4d41",
      "immutable_revision_verified": true,
      "file_sha256": {
        "model.safetensors": "53aa51172d142c89d9012cce15ae4d6cc0ca6895895114379cacb4fab128d9db",
        "tokenizer_config.json": "acb92769e8195aabd29b7b2137a9e6d6e25c476a4f15aa4355c233426c61576b",
        "special_tokens_map.json": "303df45a03609e4ead04bc3dc1536d0ab19b5358db685b6f3da123d05ec200e3",
        "config.json": "953f9c0d463486b10a6871cc2fd59f223b2c70184f49815e7efbcab5d8908b41",
        "tokenizer.json": "be50c3628f2bf5bb5e3a7f17b1f74611b2561a3a27eeab05e5aa30f411572037",
        "vocab.txt": "07eced375cec144d27c900241f3e339478dec958f92fddbc551f295c992038a3"
      },
      "local_files_only": true,
      "trust_remote_code": false,
      "verification": "pinned official-source offline byte-hash manifest"
    }
  },
  "result": "passed",
  "metrics": {
    "shared_bank": [
      {
        "seed": 42,
        "baseline": {
          "recall_at_10": 0.9,
          "ndcg_at_10": 0.7155734475958628
        },
        "test": {
          "recall_at_10": 0.9125,
          "ndcg_at_10": 0.7410298192197697
        },
        "selection": {
          "parameters": {
            "depth": 1,
            "normalize_residual": false,
            "alpha": 1
          },
          "validation": {
            "recall_at_10": 1,
            "ndcg_at_10": 0.9108720130444524
          }
        }
      },
      {
        "seed": 43,
        "baseline": {
          "recall_at_10": 0.9,
          "ndcg_at_10": 0.7155734475958628
        },
        "test": {
          "recall_at_10": 0.9125,
          "ndcg_at_10": 0.7410298192197697
        },
        "selection": {
          "parameters": {
            "depth": 1,
            "normalize_residual": false,
            "alpha": 1
          },
          "validation": {
            "recall_at_10": 1,
            "ndcg_at_10": 0.9108720130444524
          }
        }
      },
      {
        "seed": 44,
        "baseline": {
          "recall_at_10": 0.9,
          "ndcg_at_10": 0.7155734475958628
        },
        "test": {
          "recall_at_10": 0.9125,
          "ndcg_at_10": 0.7410298192197697
        },
        "selection": {
          "parameters": {
            "depth": 1,
            "normalize_residual": false,
            "alpha": 1
          },
          "validation": {
            "recall_at_10": 1,
            "ndcg_at_10": 0.9108720130444524
          }
        }
      }
    ],
    "strict_bank": [
      {
        "seed": 42,
        "baseline": {
          "recall_at_10": 0.5,
          "ndcg_at_10": 0.5
        },
        "test": {
          "recall_at_10": 0.5,
          "ndcg_at_10": 0.5
        },
        "selection": {
          "parameters": {
            "depth": 1,
            "normalize_residual": false,
            "alpha": 0.25
          },
          "validation": {
            "recall_at_10": 1,
            "ndcg_at_10": 1
          }
        }
      }
    ],
    "diagnostic_only": true,
    "promotion_eligible": false,
    "formal_comparison": false
  },
  "provenance": {
    "commit": "34148c2e29c01e9be7e4d5b8cbd09fd0f6016902",
    "artifact_path": "docs/gpu-validations/oct13-near-a30.json",
    "metrics_artifact_path": "docs/reproductions/2610.13314-near/metrics/a30-diagnostics-34148c2e.json",
    "execution_context_path": "docs/research-audits/oct14-memory-execution-context-34148c2e.json",
    "source_files_sha256": {
      "scripts/run_oct13_memory.py": "0ae99379deff465f25e42e922c72635b3646eb904f36ee031cac3f616f7d180c",
      "scripts/prepare_robotics_venv.py": "6ebc91b431f366cc614f2324d30293414140dbc4458147c873365535bdfa6c75",
      "src/auto_research/reproductions/near/__init__.py": "e7526576c299eded95c33b11c8c535d972ae4e44052d240d29f9e14f3eb18703",
      "src/auto_research/reproductions/near/adapter.py": "2c44b4430af86b7b570bb55b8d366b684ae517530aa589ebe8339f2b4db28b3d",
      "src/auto_research/reproductions/near/algorithm.py": "2058bdd67dd04dabe2482a60c234a61bcb90ad132a7e43cf40d85ef3fedd1906",
      "src/auto_research/reproductions/near/data.py": "795416bdffeb25880aeaf85b4bfd6480b8677b45cddabe37ba2e49d7bf544eb8",
      "src/auto_research/reproductions/near/experiment.py": "197e3fefad06f1f0e54533b608c5d717e57d0b6a919a105f88eee79447d94399",
      "src/auto_research/reproductions/beyond_final_embedding/__init__.py": "b61e251f3e7f6fbb26021eb9ad3996c8f8ca233175dc19b86796b695ecc44a43",
      "src/auto_research/reproductions/beyond_final_embedding/adapter.py": "3b58025b770656c4a5c2b647854a43977ce67a335abf9fda2d278ef9cd79bce9",
      "src/auto_research/reproductions/beyond_final_embedding/algorithm.py": "25ddc0d5276ca7f16d416410942a71beef721491633e0103d9b8f88a4f671622",
      "src/auto_research/reproductions/beyond_final_embedding/experiment.py": "d50cb891d364ca0c5bab3d88fa56e4aab7f4aa8f401590e49ccac32436763760"
    },
    "raw_results": [
      {
        "path": "docs/reproductions/2610.13314-near/metrics/raw/near-query-disjoint-shared-bank-100q-seed42.json",
        "sha256": "0004ee5048632ce6358b0e8a57f1726677b59de4defb6f8f7faa02a427a19878",
        "command": [
          "/tmp/oct14-memory-overlay-4496283a/bin/python",
          "scripts/run_oct13_memory.py",
          "--paper",
          "near",
          "--raw-data",
          "/tmp/auto-research-oct13-ccad3e67/public/longmemeval/longmemeval_s_cleaned.json",
          "--model-dir",
          "/tmp/auto-research-oct13-ccad3e67/public/minilm",
          "--device",
          "cuda",
          "--seed",
          "42",
          "--max-queries",
          "100",
          "--max-length",
          "128",
          "--batch-size",
          "16",
          "--adapter-epochs",
          "2",
          "--split-policy",
          "query-disjoint-shared-bank",
          "--diagnostic-only",
          "--output",
          "/tmp/auto-research-oct14-34148c2e/runs/memory-oct14/near-query-disjoint-shared-bank-100q-seed42.json"
        ]
      },
      {
        "path": "docs/reproductions/2610.13314-near/metrics/raw/near-query-disjoint-shared-bank-100q-seed43.json",
        "sha256": "323d69304a18c0e0ea2b9833543a03644fc25353325d9db2872e943f0d0c61e5",
        "command": [
          "/tmp/oct14-memory-overlay-4496283a/bin/python",
          "scripts/run_oct13_memory.py",
          "--paper",
          "near",
          "--raw-data",
          "/tmp/auto-research-oct13-ccad3e67/public/longmemeval/longmemeval_s_cleaned.json",
          "--model-dir",
          "/tmp/auto-research-oct13-ccad3e67/public/minilm",
          "--device",
          "cuda",
          "--seed",
          "43",
          "--max-queries",
          "100",
          "--max-length",
          "128",
          "--batch-size",
          "16",
          "--adapter-epochs",
          "2",
          "--split-policy",
          "query-disjoint-shared-bank",
          "--diagnostic-only",
          "--output",
          "/tmp/auto-research-oct14-34148c2e/runs/memory-oct14/near-query-disjoint-shared-bank-100q-seed43.json"
        ]
      },
      {
        "path": "docs/reproductions/2610.13314-near/metrics/raw/near-query-disjoint-shared-bank-100q-seed44.json",
        "sha256": "2c32588612766ac0225451e0e0e064a1c40bb384fb7bce534686fa8c4b417134",
        "command": [
          "/tmp/oct14-memory-overlay-4496283a/bin/python",
          "scripts/run_oct13_memory.py",
          "--paper",
          "near",
          "--raw-data",
          "/tmp/auto-research-oct13-ccad3e67/public/longmemeval/longmemeval_s_cleaned.json",
          "--model-dir",
          "/tmp/auto-research-oct13-ccad3e67/public/minilm",
          "--device",
          "cuda",
          "--seed",
          "44",
          "--max-queries",
          "100",
          "--max-length",
          "128",
          "--batch-size",
          "16",
          "--adapter-epochs",
          "2",
          "--split-policy",
          "query-disjoint-shared-bank",
          "--diagnostic-only",
          "--output",
          "/tmp/auto-research-oct14-34148c2e/runs/memory-oct14/near-query-disjoint-shared-bank-100q-seed44.json"
        ]
      },
      {
        "path": "docs/reproductions/2610.13314-near/metrics/raw/near-bank-disjoint-30q-seed42.json",
        "sha256": "23fd58177cf2743bb713ba95c7ea72ed42c1be6a6fafd6b47d05e5a861ec70f1",
        "command": [
          "/tmp/oct14-memory-overlay-4496283a/bin/python",
          "scripts/run_oct13_memory.py",
          "--paper",
          "near",
          "--raw-data",
          "/tmp/auto-research-oct13-ccad3e67/public/longmemeval/longmemeval_s_cleaned.json",
          "--model-dir",
          "/tmp/auto-research-oct13-ccad3e67/public/minilm",
          "--device",
          "cuda",
          "--seed",
          "42",
          "--max-queries",
          "30",
          "--max-length",
          "128",
          "--batch-size",
          "16",
          "--adapter-epochs",
          "2",
          "--split-policy",
          "bank-disjoint",
          "--diagnostic-only",
          "--output",
          "/tmp/auto-research-oct14-34148c2e/runs/memory-oct14/near-bank-disjoint-30q-seed42.json"
        ]
      }
    ]
  },
  "limitations": [
    "LongMemEval cleaned JSON contains official evaluation instances; source-order 100 and 30 query subsets are re-split only for a nonofficial research diagnostic, not the official benchmark test score.",
    "Shared-bank queries use fixed 60/20/20 query-ID hash splits over shared frozen unlabelled memories; no unseen-memory-bank generalization claim.",
    "The three seeds reuse the same data split and yield identical ranking metrics; they are stochastic algorithm checks, not independent dataset repetitions.",
    "Strict bank-disjoint 30-query diagnostics contain only 27 train, 1 validation and 2 test queries and are not pooled with shared-bank results.",
    "All hyperparameters/checkpoints are chosen on validation only; test labels are used only by the final scorer.",
    "MiniLM/max_length128/exact search replace original backbones/full LMEB/HNSW; no downstream QA or production result was reproduced."
  ]
}
