| { |
| "dataset": { |
| "repo_id": "lerobot/high_quality_folding", |
| "root": "lerobot/high_quality_folding", |
| "episodes": null, |
| "image_transforms": { |
| "enable": false, |
| "max_num_transforms": 3, |
| "random_order": false, |
| "tfs": { |
| "brightness": { |
| "weight": 1.0, |
| "type": "ColorJitter", |
| "kwargs": { |
| "brightness": [ |
| 0.8, |
| 1.2 |
| ] |
| } |
| }, |
| "contrast": { |
| "weight": 1.0, |
| "type": "ColorJitter", |
| "kwargs": { |
| "contrast": [ |
| 0.8, |
| 1.2 |
| ] |
| } |
| }, |
| "saturation": { |
| "weight": 1.0, |
| "type": "ColorJitter", |
| "kwargs": { |
| "saturation": [ |
| 0.5, |
| 1.5 |
| ] |
| } |
| }, |
| "hue": { |
| "weight": 1.0, |
| "type": "ColorJitter", |
| "kwargs": { |
| "hue": [ |
| -0.05, |
| 0.05 |
| ] |
| } |
| }, |
| "sharpness": { |
| "weight": 1.0, |
| "type": "SharpnessJitter", |
| "kwargs": { |
| "sharpness": [ |
| 0.5, |
| 1.5 |
| ] |
| } |
| }, |
| "affine": { |
| "weight": 1.0, |
| "type": "RandomAffine", |
| "kwargs": { |
| "degrees": [ |
| -5.0, |
| 5.0 |
| ], |
| "translate": [ |
| 0.05, |
| 0.05 |
| ] |
| } |
| } |
| } |
| }, |
| "revision": null, |
| "use_imagenet_stats": true, |
| "video_backend": "torchcodec", |
| "return_uint8": false, |
| "streaming": false |
| }, |
| "env": null, |
| "policy": { |
| "type": "vla_jepa", |
| "n_obs_steps": 1, |
| "input_features": { |
| "observation.state": { |
| "type": "STATE", |
| "shape": [ |
| 16 |
| ] |
| }, |
| "observation.images.left_wrist": { |
| "type": "VISUAL", |
| "shape": [ |
| 3, |
| 720, |
| 1280 |
| ] |
| }, |
| "observation.images.right_wrist": { |
| "type": "VISUAL", |
| "shape": [ |
| 3, |
| 720, |
| 1280 |
| ] |
| }, |
| "observation.images.base": { |
| "type": "VISUAL", |
| "shape": [ |
| 3, |
| 480, |
| 640 |
| ] |
| } |
| }, |
| "output_features": { |
| "action": { |
| "type": "ACTION", |
| "shape": [ |
| 16 |
| ] |
| } |
| }, |
| "device": "cuda", |
| "use_amp": true, |
| "use_peft": false, |
| "push_to_hub": false, |
| "repo_id": "chalkp/vla-jepa-folding", |
| "private": false, |
| "tags": [ |
| "vla-jepa", |
| "folding", |
| "bimanual", |
| "fine-tuned" |
| ], |
| "license": "apache-2.0", |
| "pretrained_path": "chalkp/vla-jepa-folding", |
| "chunk_size": 7, |
| "n_action_steps": 7, |
| "normalization_mapping": { |
| "VISUAL": "IDENTITY", |
| "STATE": "QUANTILES", |
| "ACTION": "QUANTILES" |
| }, |
| "qwen_model_name": "Qwen/Qwen3-VL-2B-Instruct", |
| "jepa_encoder_name": "facebook/vjepa2-vitl-fpc64-256", |
| "freeze_qwen": true, |
| "enable_world_model": false, |
| "reinit_modules": [ |
| "model.action_model.action_encoder", |
| "model.action_model.action_decoder", |
| "model.action_model.state_encoder" |
| ], |
| "tokenizer_padding_side": "left", |
| "prompt_template": "Your task is {instruction}. Infer the temporal dynamics from frames {actions} and produce the corresponding policy actions {e_actions}.", |
| "special_action_token": "<|action_{}|>", |
| "embodied_action_token": "<|embodied_action|>", |
| "action_dim": 16, |
| "state_dim": 16, |
| "num_action_tokens_per_timestep": 8, |
| "num_embodied_action_tokens_per_instruction": 32, |
| "num_inference_timesteps": 4, |
| "action_hidden_size": 1024, |
| "action_model_type": "DiT-B", |
| "action_num_layers": 16, |
| "action_num_heads": 12, |
| "action_attention_head_dim": 64, |
| "action_dropout": 0.2, |
| "action_num_timestep_buckets": 1000, |
| "action_noise_beta_alpha": 1.5, |
| "action_noise_beta_beta": 1.0, |
| "action_noise_s": 0.999, |
| "num_target_vision_tokens": 32, |
| "action_max_seq_len": 1024, |
| "num_video_frames": 8, |
| "predictor_depth": 12, |
| "predictor_num_heads": 8, |
| "predictor_mlp_ratio": 4.0, |
| "predictor_dropout": 0.0, |
| "world_model_loss_weight": 0.1, |
| "jepa_tubelet_size": 2, |
| "repeated_diffusion_steps": 8, |
| "resize_images_to": [ |
| 224, |
| 224 |
| ], |
| "binarize_gripper_action": false, |
| "pre_snap_gripper_action": false, |
| "clip_normalized_actions": true, |
| "gripper_dim": 7, |
| "gripper_threshold": 0.5, |
| "torch_dtype": "bfloat16", |
| "optimizer_lr": 3.75e-05, |
| "optimizer_betas": [ |
| 0.9, |
| 0.95 |
| ], |
| "optimizer_eps": 1e-08, |
| "optimizer_weight_decay": 0.01, |
| "optimizer_grad_clip_norm": 1.0, |
| "scheduler_warmup_steps": 1000, |
| "scheduler_decay_steps": 40000, |
| "scheduler_decay_lr": 2.5e-06 |
| }, |
| "reward_model": null, |
| "output_dir": "outputs/train/vla-jepa-folding", |
| "job_name": "vla-jepa-folding", |
| "resume": true, |
| "seed": 1000, |
| "cudnn_deterministic": false, |
| "num_workers": 4, |
| "batch_size": 8, |
| "prefetch_factor": 4, |
| "persistent_workers": true, |
| "steps": 40000, |
| "eval_freq": 10000, |
| "log_freq": 500, |
| "tolerance_s": 0.01, |
| "save_checkpoint": true, |
| "save_freq": 10000, |
| "use_policy_training_preset": true, |
| "optimizer": { |
| "type": "adamw", |
| "lr": 3.75e-05, |
| "weight_decay": 0.01, |
| "grad_clip_norm": 1.0, |
| "betas": [ |
| 0.9, |
| 0.95 |
| ], |
| "eps": 1e-08 |
| }, |
| "scheduler": { |
| "type": "cosine_decay_with_warmup", |
| "num_warmup_steps": 1000, |
| "num_decay_steps": 40000, |
| "peak_lr": 3.75e-05, |
| "decay_lr": 2.5e-06 |
| }, |
| "eval": { |
| "n_episodes": 0, |
| "batch_size": 0, |
| "use_async_envs": false |
| }, |
| "wandb": { |
| "enable": true, |
| "disable_artifact": true, |
| "project": "lerobot-folding", |
| "entity": null, |
| "notes": null, |
| "run_id": "ccjqh3dl", |
| "mode": "offline", |
| "add_tags": true |
| }, |
| "peft": null, |
| "sample_weighting": { |
| "type": "rabc", |
| "progress_path": "lerobot/high_quality_folding/sarm_progress.parquet", |
| "head_mode": "sparse", |
| "kappa": 0.0265, |
| "epsilon": 1e-06, |
| "extra_params": {} |
| }, |
| "rename_map": {}, |
| "checkpoint_path": "outputs/train/vla-jepa-folding/checkpoints/last" |
| } |
