| { | |
| "type": "vla0_smol", | |
| "n_obs_steps": 1, | |
| "input_features": { | |
| "observation.images.image": { | |
| "type": "VISUAL", | |
| "shape": [ | |
| 3, | |
| 256, | |
| 256 | |
| ] | |
| }, | |
| "observation.images.image2": { | |
| "type": "VISUAL", | |
| "shape": [ | |
| 3, | |
| 256, | |
| 256 | |
| ] | |
| }, | |
| "observation.state": { | |
| "type": "STATE", | |
| "shape": [ | |
| 8 | |
| ] | |
| } | |
| }, | |
| "output_features": { | |
| "action": { | |
| "type": "ACTION", | |
| "shape": [ | |
| 7 | |
| ] | |
| } | |
| }, | |
| "device": "cuda", | |
| "use_amp": true, | |
| "amp_dtype": "bf16", | |
| "push_to_hub": true, | |
| "repo_id": "olegbalakhnov/libero_vla0_final", | |
| "private": null, | |
| "tags": null, | |
| "license": null, | |
| "pretrained_path": null, | |
| "chunk_size": 8, | |
| "n_action_steps": 8, | |
| "ensemble_size": 0, | |
| "normalization_mapping": { | |
| "VISUAL": "IDENTITY", | |
| "STATE": "MIN_MAX", | |
| "ACTION": "MIN_MAX" | |
| }, | |
| "n_action_bins": 512, | |
| "n_state_bins": 512, | |
| "max_decoding_steps": 512, | |
| "use_cache": true, | |
| "optimizer_lr": 5e-05, | |
| "optimizer_betas": [ | |
| 0.9, | |
| 0.95 | |
| ], | |
| "optimizer_eps": 1e-08, | |
| "optimizer_weight_decay": 0.01, | |
| "scheduler_warmup_steps": 1000, | |
| "scheduler_decay_steps": 100000, | |
| "scheduler_decay_lr": 5e-06, | |
| "checkpoint_path": null, | |
| "vlm_checkpoint": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct", | |
| "precision": "float32", | |
| "freeze_vision_encoder": false, | |
| "do_image_splitting": false, | |
| "drop_n_last_frames": 1, | |
| "relative_actions": false, | |
| "use_state": true, | |
| "start_actions_token": 49279, | |
| "grad_clip_norm": 1.0, | |
| "action_mask_aug_per": 0.15, | |
| "action_mask_skip_per": 0.1, | |
| "crop_shape": [ | |
| 230, | |
| 230 | |
| ] | |
| } |