{
  "model_sha256": "0f282819d4238e5b7c3046d5446266dcd329d62e02139b121fbfe45fe2901d17",
  "scope": "Independent CPU PPO warm-start/fine-tuning smoke run. Same pretrained 61\u2192512\u2192256\u2192128\u219214 ELU actor and normalization; not full upstream training recipe parity or locomotion acceptance.",
  "source_sha256": "e36332d383997d51401897734cd3e79cf5038406feddb18b4d57ecfb141daa6c",
  "initial_actor_max_error": 5.960464477539062e-07,
  "export_max_error": 3.2782554626464844e-07,
  "actor_weights_changed": true,
  "updates": 10,
  "steps_per_update": 256,
  "seed": 42,
  "elapsed_s": 1.17,
  "history": [
    {
      "update": 1,
      "mean_reward": 1.2845,
      "falls": 0,
      "loss": 210.2921142578125,
      "elapsed_s": 0.1
    },
    {
      "update": 2,
      "mean_reward": 1.2891,
      "falls": 0,
      "loss": 200.2089385986328,
      "elapsed_s": 0.3
    },
    {
      "update": 3,
      "mean_reward": 1.4989,
      "falls": 0,
      "loss": 287.58587646484375,
      "elapsed_s": 0.4
    },
    {
      "update": 4,
      "mean_reward": 1.4991,
      "falls": 0,
      "loss": 266.3623352050781,
      "elapsed_s": 0.5
    },
    {
      "update": 5,
      "mean_reward": 1.4989,
      "falls": 0,
      "loss": 287.593505859375,
      "elapsed_s": 0.6
    },
    {
      "update": 6,
      "mean_reward": 1.3807,
      "falls": 0,
      "loss": 246.57568359375,
      "elapsed_s": 0.7
    },
    {
      "update": 7,
      "mean_reward": 0.7734,
      "falls": 0,
      "loss": 77.64595031738281,
      "elapsed_s": 0.8
    },
    {
      "update": 8,
      "mean_reward": 0.7701,
      "falls": 0,
      "loss": 70.28380584716797,
      "elapsed_s": 0.9
    },
    {
      "update": 9,
      "mean_reward": 0.7831,
      "falls": 0,
      "loss": 78.46422576904297,
      "elapsed_s": 1.0
    },
    {
      "update": 10,
      "mean_reward": 0.9106,
      "falls": 0,
      "loss": 93.8046646118164,
      "elapsed_s": 1.1
    }
  ],
  "checkpoint_sha256": "7e946a2c53f9eebf34b542f7d40cad206140851309dbd0c06ffc17b16885dcfe",
  "checkpoint": "micro_cat_14.onnx",
  "physical_verified": false,
  "upstream_training_recipe_parity": false
}
