| { |
| "type": "sarm", |
| "input_features": { |
| "observation.images.right_wrist": { |
| "type": "VISUAL", |
| "shape": [ |
| 480, |
| 640, |
| 3 |
| ] |
| }, |
| "observation.state": { |
| "type": "STATE", |
| "shape": [ |
| 32 |
| ] |
| } |
| }, |
| "output_features": { |
| "sparse_stage": { |
| "type": "REWARD", |
| "shape": [ |
| 13, |
| 1 |
| ] |
| }, |
| "sparse_progress": { |
| "type": "REWARD", |
| "shape": [ |
| 13, |
| 1 |
| ] |
| } |
| }, |
| "device": "cuda", |
| "pretrained_path": null, |
| "pretrained_revision": null, |
| "push_to_hub": true, |
| "repo_id": "chomeed/sarm_board_insertion", |
| "license": null, |
| "tags": null, |
| "private": null, |
| "annotation_mode": "single_stage", |
| "n_obs_steps": 8, |
| "frame_gap": 30, |
| "max_rewind_steps": 4, |
| "vision_encoder": "clip", |
| "vision_model_name": "openai/clip-vit-base-patch32", |
| "use_language": true, |
| "image_dim": 512, |
| "text_dim": 512, |
| "hidden_dim": 768, |
| "num_heads": 12, |
| "num_layers": 8, |
| "max_state_dim": 32, |
| "drop_n_last_frames": 1, |
| "batch_size": 64, |
| "clip_batch_size": 64, |
| "dropout": 0.1, |
| "stage_loss_weight": 1.0, |
| "rewind_probability": 0.8, |
| "language_perturbation_probability": 0.2, |
| "num_sparse_stages": 1, |
| "sparse_subtask_names": [ |
| "task" |
| ], |
| "sparse_temporal_proportions": [ |
| 1.0 |
| ], |
| "num_dense_stages": null, |
| "dense_subtask_names": null, |
| "dense_temporal_proportions": null, |
| "pretrained_model_path": null, |
| "image_key": "observation.images.right_wrist", |
| "state_key": "observation.state", |
| "task_override": "insert the board into the slot", |
| "normalization_mapping": { |
| "VISUAL": "IDENTITY", |
| "STATE": "MEAN_STD", |
| "LANGUAGE": "IDENTITY", |
| "REWARD": "IDENTITY" |
| } |
| } |
