{
  "model_sha256": "e68b7735f9ede610309209ac71d81a16f58724300d28687b65b7670ba060021a",
  "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": 3.5762786865234375e-07,
  "actor_weights_changed": true,
  "updates": 10,
  "steps_per_update": 256,
  "seed": 42,
  "elapsed_s": 1.06,
  "history": [
    {
      "update": 1,
      "mean_reward": 1.284,
      "falls": 0,
      "loss": 210.1680908203125,
      "elapsed_s": 0.1
    },
    {
      "update": 2,
      "mean_reward": 1.2889,
      "falls": 0,
      "loss": 200.1787109375,
      "elapsed_s": 0.2
    },
    {
      "update": 3,
      "mean_reward": 1.4989,
      "falls": 0,
      "loss": 287.5675354003906,
      "elapsed_s": 0.3
    },
    {
      "update": 4,
      "mean_reward": 1.499,
      "falls": 0,
      "loss": 266.3352355957031,
      "elapsed_s": 0.4
    },
    {
      "update": 5,
      "mean_reward": 1.4989,
      "falls": 0,
      "loss": 287.5653076171875,
      "elapsed_s": 0.5
    },
    {
      "update": 6,
      "mean_reward": 1.3802,
      "falls": 0,
      "loss": 246.501220703125,
      "elapsed_s": 0.6
    },
    {
      "update": 7,
      "mean_reward": 0.8442,
      "falls": 0,
      "loss": 93.20362091064453,
      "elapsed_s": 0.7
    },
    {
      "update": 8,
      "mean_reward": 0.8212,
      "falls": 0,
      "loss": 79.9521484375,
      "elapsed_s": 0.8
    },
    {
      "update": 9,
      "mean_reward": 0.8564,
      "falls": 0,
      "loss": 93.862060546875,
      "elapsed_s": 0.9
    },
    {
      "update": 10,
      "mean_reward": 0.9656,
      "falls": 0,
      "loss": 105.16844940185547,
      "elapsed_s": 1.0
    }
  ],
  "checkpoint_sha256": "fc312f9043f6ae4a78f63bb3f48738102aae24f754cca2454d3f38c116ed5845",
  "checkpoint": "micro_x_14.onnx",
  "physical_verified": false,
  "upstream_training_recipe_parity": false
}