naavox/multitask-dit-2

1{2    "dataset": {3        "repo_id": "naavox/laundry_grasp_one_cam",4        "root": null,5        "episodes": null,6        "image_transforms": {7            "enable": true,8            "max_num_transforms": 3,9            "random_order": false,10            "tfs": {11                "brightness": {12                    "weight": 1.0,13                    "type": "ColorJitter",14                    "kwargs": {15                        "brightness": [16                            0.8,17                            1.218                        ]19                    }20                },21                "contrast": {22                    "weight": 1.0,23                    "type": "ColorJitter",24                    "kwargs": {25                        "contrast": [26                            0.8,27                            1.228                        ]29                    }30                },31                "saturation": {32                    "weight": 1.0,33                    "type": "ColorJitter",34                    "kwargs": {35                        "saturation": [36                            0.5,37                            1.538                        ]39                    }40                },41                "hue": {42                    "weight": 1.0,43                    "type": "ColorJitter",44                    "kwargs": {45                        "hue": [46                            -0.05,47                            0.0548                        ]49                    }50                },51                "sharpness": {52                    "weight": 1.0,53                    "type": "SharpnessJitter",54                    "kwargs": {55                        "sharpness": [56                            0.5,57                            1.558                        ]59                    }60                },61                "affine": {62                    "weight": 1.0,63                    "type": "RandomAffine",64                    "kwargs": {65                        "degrees": [66                            -5.0,67                            5.068                        ],69                        "translate": [70                            0.05,71                            0.0572                        ]73                    }74                }75            }76        },77        "revision": null,78        "use_imagenet_stats": true,79        "video_backend": "torchcodec",80        "return_uint8": false,81        "streaming": false82    },83    "env": null,84    "policy": {85        "type": "multi_task_dit",86        "n_obs_steps": 2,87        "input_features": {88            "observation.state": {89                "type": "STATE",90                "shape": [91                    1792                ]93            },94            "observation.images.gripper_camera": {95                "type": "VISUAL",96                "shape": [97                    3,98                    384,99                    384100                ]101            }102        },103        "output_features": {104            "action": {105                "type": "ACTION",106                "shape": [107                    5108                ]109            }110        },111        "device": "cuda",112        "use_amp": false,113        "use_peft": false,114        "push_to_hub": true,115        "repo_id": "naavox/multitask-dit-2",116        "private": null,117        "tags": null,118        "license": null,119        "pretrained_path": null,120        "horizon": 64,121        "n_action_steps": 32,122        "objective": "flow_matching",123        "noise_scheduler_type": "DDPM",124        "num_train_timesteps": 100,125        "beta_schedule": "squaredcos_cap_v2",126        "beta_start": 0.0001,127        "beta_end": 0.02,128        "prediction_type": "epsilon",129        "clip_sample": true,130        "clip_sample_range": 1.0,131        "num_inference_steps": null,132        "sigma_min": 0.0,133        "num_integration_steps": 100,134        "integration_method": "euler",135        "timestep_sampling_strategy": "beta",136        "timestep_sampling_s": 0.999,137        "timestep_sampling_alpha": 1.5,138        "timestep_sampling_beta": 1.0,139        "hidden_dim": 512,140        "num_layers": 6,141        "num_heads": 8,142        "dropout": 0.1,143        "use_positional_encoding": false,144        "timestep_embed_dim": 256,145        "use_rope": true,146        "rope_base": 10000.0,147        "vision_encoder_name": "openai/clip-vit-base-patch16",148        "use_separate_rgb_encoder_per_camera": false,149        "vision_encoder_lr_multiplier": 0.1,150        "image_resize_shape": [151            256,152            256153        ],154        "image_crop_shape": [155            224,156            224157        ],158        "image_crop_is_random": true,159        "text_encoder_name": "openai/clip-vit-base-patch16",160        "tokenizer_max_length": 77,161        "tokenizer_padding": "max_length",162        "tokenizer_padding_side": "right",163        "tokenizer_truncation": true,164        "normalization_mapping": {165            "VISUAL": "MEAN_STD",166            "STATE": "MIN_MAX",167            "ACTION": "MIN_MAX"168        },169        "optimizer_lr": 2e-05,170        "optimizer_betas": [171            0.95,172            0.999173        ],174        "optimizer_eps": 1e-08,175        "optimizer_weight_decay": 0.0,176        "scheduler_name": "cosine",177        "scheduler_warmup_steps": 0,178        "do_mask_loss_for_padding": false,179        "drop_n_last_frames": 31180    },181    "reward_model": null,182    "output_dir": "outputs/multitask_dit_training_2",183    "job_name": "multi_task_dit",184    "resume": false,185    "seed": 1000,186    "cudnn_deterministic": false,187    "num_workers": 4,188    "batch_size": 90,189    "prefetch_factor": 4,190    "persistent_workers": true,191    "steps": 30000,192    "eval_freq": 20000,193    "log_freq": 200,194    "tolerance_s": 0.0001,195    "save_checkpoint": true,196    "save_freq": 20000,197    "use_policy_training_preset": true,198    "optimizer": {199        "type": "adam",200        "lr": 2e-05,201        "weight_decay": 0.0,202        "grad_clip_norm": 10.0,203        "betas": [204            0.95,205            0.999206        ],207        "eps": 1e-08208    },209    "scheduler": {210        "type": "diffuser",211        "num_warmup_steps": 0,212        "name": "cosine"213    },214    "eval": {215        "n_episodes": 50,216        "batch_size": 16,217        "use_async_envs": true218    },219    "wandb": {220        "enable": false,221        "disable_artifact": false,222        "project": "lerobot",223        "entity": null,224        "notes": null,225        "run_id": null,226        "mode": null,227        "add_tags": true228    },229    "peft": null,230    "sample_weighting": null,231    "rename_map": {},232    "checkpoint_path": null233}