{
 "aggregates": {
  "Belebele [confirmation]": {
   "echo_acc": 0.9154,
   "echo_correct": 119,
   "fable_acc": 0.9538,
   "fable_correct": 124,
   "fable_valid_n": 130,
   "n": 130
  },
  "GPQA Diamond [full paired development (n=100)]": {
   "echo_acc": 0.94,
   "echo_correct": 94,
   "fable_acc": 0.94,
   "fable_correct": 94,
   "fable_valid_n": 100,
   "n": 100
  },
  "GPQA Diamond [held-out partial comparator (n=41)]": {
   "echo_acc": 0.9024,
   "echo_correct": 37,
   "fable_acc": 0.8947,
   "fable_correct": 17,
   "fable_valid_n": 19,
   "n": 41
  },
  "GPQA Diamond [post-seal availability diagnostic (n=19)]": {
   "echo_acc": 0.9474,
   "echo_correct": 18,
   "fable_acc": 0.8947,
   "fable_correct": 17,
   "fable_valid_n": 19,
   "n": 19
  },
  "Global-MMLU [confirmation]": {
   "echo_acc": 0.8868,
   "echo_correct": 94,
   "fable_acc": 0.9434,
   "fable_correct": 100,
   "fable_valid_n": 106,
   "n": 106
  },
  "HumanEval+ [confirmation]": {
   "echo_acc": 0.9355,
   "echo_correct": 29,
   "fable_acc": 0.9355,
   "fable_correct": 29,
   "fable_valid_n": 31,
   "n": 31
  },
  "MATH-500 [canonical (n=249)]": {
   "echo_acc": 0.988,
   "echo_correct": 246,
   "fable_acc": 0.988,
   "fable_correct": 246,
   "fable_valid_n": 249,
   "n": 249
  },
  "MMLU-Pro [confirmation]": {
   "echo_acc": 0.8786,
   "echo_correct": 123,
   "fable_acc": 0.9,
   "fable_correct": 126,
   "fable_valid_n": 140,
   "n": 140
  },
  "MedMCQA [confirmation]": {
   "echo_acc": 0.9231,
   "echo_correct": 84,
   "fable_acc": 0.9231,
   "fable_correct": 84,
   "fable_valid_n": 91,
   "n": 91
  }
 },
 "correctness_policy": "attested stored grades primary; deterministic text grader as cross-check/fallback; code rows use stored grades only",
 "generated_at": "2026-07-18T23:34:27Z",
 "name": "Echo Eval Observatory data bundle",
 "note": "Hashes bind this bundle to declared artifacts. Source files remain internal, so independent source verification is unavailable; hashes do not establish preregistration or rule out selection bias.",
 "row_count": 907,
 "sources": [
  {
   "role": "GPQA comparator availability and denominator audit",
   "sha256": "a65efdfb3bd2c1b79d07140f9d8af894287a60286878134fe4be1864e93d5801"
  },
  {
   "role": "Echo per-row outputs",
   "sha256": "84f83e2aba0ab58306cd92667f89f99947e1bb8cd3f96bd8bb7d08d204d9cd89"
  },
  {
   "role": "Fable per-row outputs",
   "sha256": "9dee2d66c77d7d8a5fb35348c84942463c7bc6c908fdfbf4e5a7e24069dcacee"
  },
  {
   "role": "frozen fixture (GPQA Diamond full paired n=100)",
   "sha256": "4904a7197fc84e474f0d04e32a07fc7f0061af756ac14857f544d6a05d6ea4ce"
  },
  {
   "role": "gpqa_fresh_echo",
   "sha256": "096182172d98a4b61614d16b6a0f06d898ee5f24cf2e1d19a8d6680e95a97570"
  },
  {
   "role": "gpqa_fresh_fable",
   "sha256": "6e266a940cb6d9c922c84b49b2aeafd042df7b40faec11fa468345e4e66253e1"
  },
  {
   "role": "gpqa_fresh_fixture",
   "sha256": "c67e4c1c2eded5d702eebbfddb7c62009a5e05e707abd97ab2dcc65c824e9c0b"
  },
  {
   "role": "gpqa_fresh_gold",
   "sha256": "81e6843633f20dad89128c28e46183a02048f110eec43cc8544dc16d6ba0f5f0"
  },
  {
   "role": "gpqa_fresh_grades",
   "sha256": "580a7b6ad559a57ff656f4081ee225cbcb2ce7e33908280f9653249aeb529e92"
  },
  {
   "role": "GPQA paired aggregate and source-hash report",
   "sha256": "78d75a79dfa9f294f0de2ff2d16511b3c883d18057dfd6f9fa077e415d76e9f6"
  },
  {
   "role": "attested GPQA per-row grades",
   "sha256": "d7575b3e4d52969f126a94463f10fd72cb55118c6673c7fdc37723758853d04b"
  },
  {
   "role": "pre-grading GPQA cohort and policy freeze",
   "sha256": "9872e7229a90c081aaba129fe961b0a9c2d1f0c6cfe0270be7712ced9f461a6c"
  },
  {
   "role": "math_verify",
   "sha256": "6f12ac755d1082a9da3ee4ee67fa6a35f90f32f11d0b22fd57856b7a415d9261"
  },
  {
   "role": "Echo per-row outputs",
   "sha256": "798c7e7000bb032f460193696cb9826e8367466ba7143636488a085348551b8f"
  },
  {
   "role": "Fable per-row outputs",
   "sha256": "3f86be26bdc30ad188c4e1e09e79012c94e37c2584d648a88586911ef6565f21"
  },
  {
   "role": "frozen fixture (multi-benchmark cohort)",
   "sha256": "e4ec4ec55b0c3c9a8d9ab37cdf66a2a8e113c09b686e2324d5c4ba56084b4801"
  },
  {
   "role": "attested grade file (Echo)",
   "sha256": "b088696cbc2a5b3f91fd0f66b8107a222e3a5fa52faaa7671b015713bd398fc6"
  },
  {
   "role": "attested grade file (Fable)",
   "sha256": "60830d6d024ab455b564f17ba8a235a794e7869abf57bf859102edb38d9665d8"
  }
 ]
}