training_artifact.json

{
  "trainer_kind": "grpo",
  "base_model": "openai/gpt-oss-20b",
  "renderer_name": "gpt_oss_medium_reasoning",
  "checkpoint_uri": "tinker://74ca18ef-36ae-5a4d-9aab-21ac220f42dd:train:0/sampler_weights/final",
  "token_usage": {
    "sampling_input_tokens": 3053072,
    "sampling_output_tokens": 825595,
    "num_examples": 2322,
    "num_steps": 20
  },
  "trainer_config_snapshot": {
    "kind": "grpo",
    "base_model": "openai/gpt-oss-20b",
    "renderer_name": "gpt_oss_medium_reasoning",
    "learning_rate": 0.00004,
    "group_size": 8,
    "groups_per_batch": 8,
    "num_iters": 20,
    "save_every": 5,
    "max_tokens": 8192,
    "lora_rank": 32,
    "temperature": 1.0,
    "reward": {
      "kind": "correctness"
    }
  }
}