Upload Piper adaptive B-spline checkpoint_validation.json

models/piper-pick-white-block-25hz-qwenpi-v3-adaptive-left-exclude-gripper-native-controls13-gb32-50k-20260905/checkpoint_validation.json

@@ -0,0 +1,24 @@
+{
+  "strict_reload": true,
+  "checkpoint_bytes": 10920597069,
+  "checkpoint_sha256": "23656dc0823fdbe8b1f285b83d451fdbb930f6f1789c1f39d4ab98729e64f9f6",
+  "remote_lfs_size_matches": false,
+  "remote_lfs_sha256_matches": false,
+  "framework": "QwenPI_v3_AdaptiveBSpline",
+  "training_objective": "native_flow_matching_velocity_mse",
+  "l1_loss_used": false,
+  "parameter_shape_per_observation": [13, 8],
+  "parameter_layout": "[control_dim_0 ... control_dim_6, duration_steps]",
+  "decoded_action_shape": "[predicted_native_horizon, 7]",
+  "metadata_action_chunk_size": null,
+  "metadata_variable_action_chunk": true,
+  "metadata_max_native_horizon": 64,
+  "metadata_frequency_hz": 25.0,
+  "predicted_native_horizon_range": [10, 49],
+  "target_native_horizon_range": [10, 48],
+  "full_open_loop_evaluation_passed": true,
+  "evaluated_observations": 17141,
+  "evaluated_episodes": 52,
+  "representation_arm_rmse_improvement_percent_vs_uniform_common_prefix": 20.174699988116174,
+  "policy_arm_rmse_improvement_percent_vs_uniform_common_prefix": -6.71582066910652
+}