{
  "domain": "computer-science",
  "image": "python:3.12-slim",
  "data": "FineWeb 10B tokens with the GPT-2 tokenizer, in a fixed order; the validation set is the first 10M tokens.",
  "benchmark": {
    "task": "Budgeted language-model training",
    "metric": {
      "name": "minutes to 3.28",
      "better": "lower"
    },
    "guardrails": [
      "Reaches 3.28 cross-entropy on the FineWeb validation set",
      "One run on the pinned 8×H100 runner, with no warm caches",
      "Same tokenizer, data and order; no external weights or data"
    ],
    "baseline": {
      "name": "llm.c GPT-2 (124M)",
      "score": 45
    }
  }
}