{
  "actual_author_reward_function_outputs": [
    {
      "id": "math-ok",
      "math_reward": 1,
      "format_reward": 1
    },
    {
      "id": "math-wrong",
      "math_reward": 0,
      "format_reward": 1
    },
    {
      "id": "caption",
      "math_reward": null,
      "format_reward": 1
    },
    {
      "id": "format-missing",
      "math_reward": 0,
      "format_reward": 0
    }
  ],
  "none_is_not_zero": true,
  "trl_grpo_trainer_executed": false,
  "evidence_kind": "author input/reference computation, not a target product or model run",
  "paid_or_network_model_calls": 0
}
