| { | |
| "type": "cascaded_flow", | |
| "n_obs_steps": 2, | |
| "normalization_mapping": { | |
| "VISUAL": "MEAN_STD", | |
| "STATE": "MIN_MAX", | |
| "ACTION": "MIN_MAX" | |
| }, | |
| "input_features": { | |
| "observation.images.primary": { | |
| "type": "VISUAL", | |
| "shape": [ | |
| 3, | |
| 256, | |
| 256 | |
| ] | |
| }, | |
| "observation.images.tactile": { | |
| "type": "VISUAL", | |
| "shape": [ | |
| 3, | |
| 256, | |
| 256 | |
| ] | |
| }, | |
| "observation.images.wrist": { | |
| "type": "VISUAL", | |
| "shape": [ | |
| 3, | |
| 256, | |
| 256 | |
| ] | |
| }, | |
| "observation.state.cartesian": { | |
| "type": "STATE", | |
| "shape": [ | |
| 6 | |
| ] | |
| }, | |
| "observation.state.gripper": { | |
| "type": "STATE", | |
| "shape": [ | |
| 1 | |
| ] | |
| }, | |
| "observation.state.joints": { | |
| "type": "STATE", | |
| "shape": [ | |
| 7 | |
| ] | |
| }, | |
| "observation.state.sensors_ft_sensor": { | |
| "type": "STATE", | |
| "shape": [ | |
| 6 | |
| ] | |
| }, | |
| "observation.state.sensors_tactile_sensor": { | |
| "type": "STATE", | |
| "shape": [ | |
| 15 | |
| ] | |
| }, | |
| "observation.state.target": { | |
| "type": "STATE", | |
| "shape": [ | |
| 6 | |
| ] | |
| }, | |
| "observation.state": { | |
| "type": "STATE", | |
| "shape": [ | |
| 41 | |
| ] | |
| } | |
| }, | |
| "output_features": { | |
| "action": { | |
| "type": "ACTION", | |
| "shape": [ | |
| 7 | |
| ] | |
| } | |
| }, | |
| "device": "cuda", | |
| "use_amp": false, | |
| "push_to_hub": true, | |
| "repo_id": "danielsanjosepro/cascaded_flow_stack_v2", | |
| "private": null, | |
| "tags": null, | |
| "license": null, | |
| "slow_policy_config": { | |
| "crop_shape": [ | |
| 224, | |
| 224 | |
| ], | |
| "use_separate_rgb_encoder_per_camera": true, | |
| "horizon": 16, | |
| "n_action_steps": 6, | |
| "n_obs_steps": 2, | |
| "optimizer_lr": 0.0001, | |
| "training_noise_sampling": "uniform", | |
| "features_to_exclude": [ | |
| "observation.images.tactile" | |
| ] | |
| }, | |
| "fast_policy_config": { | |
| "crop_shape": [ | |
| 224, | |
| 224 | |
| ], | |
| "use_separate_rgb_encoder_per_camera": true, | |
| "horizon": 8, | |
| "n_action_steps": 1, | |
| "n_obs_steps": 2, | |
| "frequency_embedding_dim": 128, | |
| "hidden_dim": 256, | |
| "num_blocks": 6, | |
| "num_heads": 16, | |
| "dropout": 0.1, | |
| "dim_feedforward": 2048, | |
| "features_to_exclude": [ | |
| "observation.images.primary", | |
| "observation.images.wrist", | |
| "observation.images.tactile" | |
| ] | |
| }, | |
| "action_batch_size": 20, | |
| "sampling_strategy": "stochastic", | |
| "sampling_temperature": 1.0, | |
| "smooth_slow_policy_output": false, | |
| "apply_slow_policy_output": false, | |
| "use_image_features_for_fast_policy": true, | |
| "training_stage": "fast_policy", | |
| "slow_policy_type": "ditmeanflow" | |
| } |