Instructions to use ETHrobotlearning/molmoact2-task3-TOY-clean-step2000 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LeRobot
How to use ETHrobotlearning/molmoact2-task3-TOY-clean-step2000 with LeRobot:
- Notebooks
- Google Colab
- Kaggle
| { | |
| "type": "molmoact2", | |
| "n_obs_steps": 1, | |
| "input_features": { | |
| "observation.state": { | |
| "type": "STATE", | |
| "shape": [ | |
| 6 | |
| ] | |
| }, | |
| "observation.images.front": { | |
| "type": "VISUAL", | |
| "shape": [ | |
| 3, | |
| 480, | |
| 640 | |
| ] | |
| } | |
| }, | |
| "output_features": { | |
| "action": { | |
| "type": "ACTION", | |
| "shape": [ | |
| 6 | |
| ] | |
| } | |
| }, | |
| "device": "cuda", | |
| "use_amp": false, | |
| "use_peft": false, | |
| "push_to_hub": true, | |
| "repo_id": "ETHrobotlearning/molmoact2-task3-TOY-clean-step2000", | |
| "private": null, | |
| "tags": null, | |
| "license": null, | |
| "pretrained_path": null, | |
| "checkpoint_path": "allenai/MolmoAct2-SO100_101", | |
| "checkpoint_revision": null, | |
| "checkpoint_force_download": false, | |
| "trust_remote_code": true, | |
| "chunk_size": 10, | |
| "n_action_steps": 10, | |
| "action_mode": "continuous", | |
| "inference_action_mode": null, | |
| "discrete_action_tokenizer": "allenai/MolmoAct2-FAST-Tokenizer", | |
| "discrete_generation_max_steps": null, | |
| "norm_tag": null, | |
| "setup_type": "single SO-100/SO-101 arm with one front RGB camera", | |
| "control_mode": "absolute joint pose", | |
| "image_keys": [ | |
| "observation.images.front" | |
| ], | |
| "normalize_language": true, | |
| "add_setup_tokens": true, | |
| "add_control_tokens": true, | |
| "normalize_gripper": false, | |
| "num_state_tokens": 256, | |
| "max_sequence_length": null, | |
| "expected_max_action_dim": 32, | |
| "num_flow_timesteps": 8, | |
| "flow_matching_cutoff": 1.0, | |
| "flow_matching_time_offset": 0.001, | |
| "flow_matching_time_scale": 0.999, | |
| "flow_matching_beta_alpha": 1.0, | |
| "flow_matching_beta_beta": 1.5, | |
| "num_inference_steps": null, | |
| "mask_action_dim_padding": true, | |
| "enable_inference_cuda_graph": true, | |
| "per_episode_seed": false, | |
| "eval_seed": null, | |
| "rtc_config": null, | |
| "enable_lora_vlm": true, | |
| "lora_rank": 64, | |
| "lora_alpha": 16, | |
| "lora_dropout": 0.05, | |
| "lora_bias": "none", | |
| "enable_lora_action_expert": false, | |
| "enable_knowledge_insulation": false, | |
| "freeze_embedding": true, | |
| "train_action_expert_only": false, | |
| "gradient_checkpointing": true, | |
| "model_dtype": "bfloat16", | |
| "softmax_auxiliary_loss": true, | |
| "softmax_auxiliary_loss_scale": 0.0001, | |
| "discrete_loss_token_weighting": "root_subsegments_root_tokens", | |
| "optimizer_lr": 1e-05, | |
| "optimizer_vit_lr": 5e-06, | |
| "optimizer_connector_lr": 5e-06, | |
| "optimizer_action_expert_lr": 5e-05, | |
| "optimizer_betas": [ | |
| 0.9, | |
| 0.95 | |
| ], | |
| "optimizer_eps": 1e-06, | |
| "optimizer_weight_decay": 0.0, | |
| "optimizer_grad_clip_norm": 1.0, | |
| "scheduler_warmup_steps": 200, | |
| "scheduler_decay_steps": null, | |
| "scheduler_decay_lr": 1e-06, | |
| "normalization_mapping": { | |
| "VISUAL": "IDENTITY", | |
| "STATE": "QUANTILES", | |
| "ACTION": "QUANTILES" | |
| }, | |
| "dataset_feature_names": { | |
| "action": [ | |
| "shoulder_pan.pos", | |
| "shoulder_lift.pos", | |
| "elbow_flex.pos", | |
| "wrist_flex.pos", | |
| "wrist_roll.pos", | |
| "gripper.pos" | |
| ], | |
| "observation.state": [ | |
| "shoulder_pan.pos", | |
| "shoulder_lift.pos", | |
| "elbow_flex.pos", | |
| "wrist_flex.pos", | |
| "wrist_roll.pos", | |
| "gripper.pos" | |
| ] | |
| } | |
| } |