{
  "documents": 807,
  "reference_sha256": {
    "/data/calib/imatrix_calib.txt": "27e416716dd0a6d4bcb931a1ab32dd38a711a78fae887d24ed0ad04fe00ef5d8",
    "/data/calib/wiki.test.raw": "173c87a53759e0201f33e0ccf978e510c2042d7f2cb78229d9a50d79b9e7dd08",
    "/data/calib/imatrix-input-0.txt": "35b8690af93aff8d02f37081fa37447de9cd6bc68ff7eb349c4a0b7e3d86ba31",
    "/data/calib/imatrix-input-1.txt": "dd369c73ec39fcfc114e9b7975652d3869c295b37385b6dfbeb84534ca297f87",
    "/data/calib/imatrix-input-2.txt": "baae4e9cf85f18be0fd2660c7b6ddbc0490f6917e91ec9d2f418e1c23692156b",
    "/data/calib/imatrix-input-3.txt": "ca54fd14852eefe7e015ae8c267381f0715b81d5b1b7bbdd65fb5879f3343e2e"
  },
  "documents_metadata_sha256": "26a77a21f73d1d37082f494df46f813a785454265b74c1facf4441f3db07a0a4",
  "domain_sha256": {
    "prose_c4": "4a41458d30ddcc39e55a0fbc2913f88c74a25ada5ae4005e27234b1881b382e9",
    "code_codeparrot": "fba3cf09746177fc6518c8dfd444f8d6f7ef0ccde79fb8388bd60806fae49c01",
    "math_gsm8k": "e23dc6264fb1fa035de118753afb2a4ca7469c792e7a467a64c541152b60ca14",
    "multilingual_mc4": "7933252e477bf0f356cddcdb8cea8595f2c7199c34a05cdf590bd03d3a0d549c"
  },
  "numeric_normalization": "Replace digit runs with #; casefold",
  "passes": {
    "word5": {
      "window": 256,
      "stride": 128,
      "minimum_shared": 20,
      "containment_threshold": 0.5
    },
    "char25": {
      "window": 1024,
      "stride": 512,
      "minimum_shared": 50,
      "containment_threshold": 0.6
    }
  },
  "limitation": "Lexical similarity screen against reference windows; not semantic deduplication or source-document-ID proof. No model scores read by this audit."
}