{
  "arxiv": "2510.14961v1",
  "attempt_id": "534db42c-5b16-4f00-9a7d-a47056fc9dd4",
  "claims": {
    "claim_1": {
      "evidence_status": "partial",
      "sha256": "d0da87ee16f7485d3dff369e7465f66299c55ac003a54e1cf8c00b3a0ad8b265",
      "text": "The sampler decodes new tokens every forward pass while refining latent states for those tokens in parallel through recurrent depth (Section 3.1)."
    },
    "claim_2": {
      "evidence_status": "unavailable",
      "sha256": "2e15221c8b5516b0ab705e29a3d7c5d924ed5f0187c970a0caf60a1402757804",
      "text": "The paper proves the sampler is strictly more expressive than baseline autoregressive generation under the same time budget on modern hardware (Theorem 4.2)."
    }
  },
  "commands": {
    "evidence_generation": "uv run --locked --project submissions/efficient-parallel-samplers-for-recurrent-depth-models python -m recurrent_sampler_repro.evidence --project-root submissions/efficient-parallel-samplers-for-recurrent-depth-models",
    "test_suite": "uv run --locked --project submissions/efficient-parallel-samplers-for-recurrent-depth-models python -m pytest submissions/efficient-parallel-samplers-for-recurrent-depth-models/tests"
  },
  "inputs": {
    "ATTRIBUTION.md": {
      "license": "CC-BY-4.0",
      "sha256": "79775b50c72988b90eae75ef87e9d4df9dbd0bfceefaed60b398656a88d8a735",
      "verified": true
    },
    "LICENSE": {
      "license": "Apache-2.0",
      "sha256": "bc6c264d8ba4450599cf95c4699c6b82142f32ca1ecd91011c17b50a5a36a2f5",
      "verified": true
    },
    "arxiv_submission.tex": {
      "sha256": "cdc058830d1e51f631e4fb8d1f2de0b79de91670fd4111646fe624f8c258d3b8",
      "verified": true
    },
    "pdf_sha256": "74e7985abe41ee2a75914a65e3778a15353fb0c0964d6ea34e7bfeb1f18312c8",
    "raven_modeling_minimal.py": {
      "git_blob": "0e83a0766644df9113a8923f43350c6a1b5a182c",
      "sha256": "18fcacd53fb5696a76c0d3bda44480f2f3900aa9659c137a08962c593a9a9e42",
      "verified": true
    },
    "source_archive_sha256": "60a795d123a2d2d642971834b6e0cba6dda80b5dfcd539f78d01639582d9c41d"
  },
  "outputs": {
    "REPORT.md": {
      "sha256": "31e26557097bb5ef94dcad02fd4258c038a81530d7ebbf6d8d168fbc3d675a67"
    },
    "evidence/REPORT.md": {
      "sha256": "31e26557097bb5ef94dcad02fd4258c038a81530d7ebbf6d8d168fbc3d675a67"
    },
    "evidence/claim-1-wavefront.json": {
      "sha256": "03760fa304a2fac5de0b7ffa8d2a41351c251635f0cbda0c1d3f1b2efe9d7d52"
    },
    "evidence/claim-2-theorem-audit.json": {
      "sha256": "9677bc7f4948e8a12621f80191ebe5c40487ae457035cd50cbebaee0b079bd17"
    },
    "evidence/manifest.json": {
      "sha256": null,
      "unhashed_reason": "Self-referential manifest copy"
    },
    "evidence/results.json": {
      "sha256": "487e2994ca3010a6289d6b48b508b02d6c00b11fadb2d356a2f0329b17e06396"
    },
    "index.html": {
      "sha256": "beaf88f0fb10a020594bb5baf6b33d531728028f6537942f7cd5d65b1eb71bc9"
    },
    "poster.html": {
      "sha256": "beaf88f0fb10a020594bb5baf6b33d531728028f6537942f7cd5d65b1eb71bc9"
    },
    "space/README.md": {
      "sha256": "d8df6642131d5defdd20c240c6dd88b93c2e33f21d545e398381d30e9a0a825e"
    },
    "space/REPORT.md": {
      "sha256": "31e26557097bb5ef94dcad02fd4258c038a81530d7ebbf6d8d168fbc3d675a67"
    },
    "space/evidence/claim-1-wavefront.json": {
      "sha256": "03760fa304a2fac5de0b7ffa8d2a41351c251635f0cbda0c1d3f1b2efe9d7d52"
    },
    "space/evidence/claim-2-theorem-audit.json": {
      "sha256": "9677bc7f4948e8a12621f80191ebe5c40487ae457035cd50cbebaee0b079bd17"
    },
    "space/evidence/manifest.json": {
      "sha256": null,
      "unhashed_reason": "Self-referential manifest copy"
    },
    "space/evidence/results.json": {
      "sha256": "487e2994ca3010a6289d6b48b508b02d6c00b11fadb2d356a2f0329b17e06396"
    },
    "space/index.html": {
      "sha256": "beaf88f0fb10a020594bb5baf6b33d531728028f6537942f7cd5d65b1eb71bc9"
    },
    "space/poster.html": {
      "sha256": "beaf88f0fb10a020594bb5baf6b33d531728028f6537942f7cd5d65b1eb71bc9"
    }
  },
  "paper_id": "h7WBYYJF1Q",
  "provenance_token": "arxiv:2510.14961v1+pdf-sha256:74e7985abe41ee2a75914a65e3778a15353fb0c0964d6ea34e7bfeb1f18312c8+source-sha256:60a795d123a2d2d642971834b6e0cba6dda80b5dfcd539f78d01639582d9c41d+github:seal-rg/recurrent-pretraining@1ea7220ec7eb42d13e89db0663df254d0bcdc28e+git-blob:recpre/raven_modeling_minimal.py@0e83a0766644df9113a8923f43350c6a1b5a182c",
  "python_requirement": ">=3.10"
}