{
  "seed": 1147,
  "per_source": {
    "WithinUsAI/GOD_Coder_100k": {
      "raw": 100000,
      "unique": 100000
    },
    "WithinUsAI/Elite_GOD_Coder_100k": {
      "raw": 100000,
      "unique": 100000
    },
    "WithinUsAI/Omega_Genesis_Coder_100k": {
      "raw": 100000,
      "unique": 100000
    },
    "WithinUsAI/python_GOD_coder_100k": {
      "raw": 100000,
      "unique": 99498
    },
    "WithinUsAI/GOD_Coder_Complete_DataSet": {
      "raw": 175000,
      "unique": 173860
    },
    "WithinUsAI/Royal_Ghost_Coder_501k": {
      "raw": 490000,
      "unique": 422124
    }
  },
  "duplicates_removed": 69518,
  "malformed_skipped": 0,
  "files": {
    "train.jsonl": {
      "rows": 993482,
      "sha256": "f6431d336856711a2318a5fa4c6b1ddcc9d1b007e0e274c4dc3304778b83ef37",
      "bytes": 1766010503
    },
    "validation.jsonl": {
      "rows": 2000,
      "sha256": "f63487515c57439f27dd9e848719dd60f1a0de229a23e3215913c467d4eeb9b0",
      "bytes": 3500556
    }
  },
  "corpus_card": "AGA-Code clean pretraining corpus v1.0.0 \u2014 six WithinUsAI coder datasets (4 gunzipped, messages/instruction-response rendered), SHA-256 deduped, seed-1147 shuffled, 2000-row held-out validation."
}