evals/chatcore.json

{
  "stage": "sft",
  "step": 42,
  "total_training_flops": 2.3204581023414642e+20,
  "stage_training_flops": 1.0107782648656036e+18,
  "inherited_parent_flops": 2.3103503196928082e+20,
  "cumulative_pipeline_training_flops": 2.3204581023414642e+20,
  "results": {
    "ARC-Easy": 0.24579124579124578,
    "ARC-Challenge": 0.23037542662116042,
    "MMLU": 0.23137729668138443,
    "GSM8K": 0.0,
    "HumanEval": 0.0
  },
  "chatcore_metric": -0.011321608241655831,
  "chatcore_suite": {
    "name": "karpathy",
    "tasks": [
      "ARC-Easy",
      "ARC-Challenge",
      "MMLU",
      "GSM8K",
      "HumanEval"
    ],
    "max_generative_problems": null,
    "generative_answer_format": null
  },
  "complete": true
}