{
  "base_model": "Qwen/Qwen2.5-0.5B-Instruct",
  "data": "data/mixed_train.jsonl",
  "steps": 10800,
  "epochs": 1,
  "tasks": [
    "all"
  ],
  "stratified": true,
  "batch_size": 4,
  "final_avg_loss": 1.1945149920370293,
  "samples_by_gym": {
    "chess": 10800,
    "ticket_dungeon": 10800,
    "resource_allocator": 10800,
    "debate_judge": 10800
  }
}