seelanrohan/multi_task_dit_towel_v1

multi_task_dit Training Run

Summary

• Model: multi_task_dit
• Policy type: multi_task_dit
• Dataset: ["['small_towel_folding_attempt_3_truncated_384'", "'small_towel_folding_attempt_4_truncated_384'", "'small_towel_folding_attempt_5_truncated_384']"]
• Steps: 80k, batch_size=200
• Optimizer: Adam lr=2e-5, diffuser schedule w/ 0 warmup steps
• Checkpoints: every 10k
• push_to_hub: false
• wandb: enabled (project=lerobot)

Full Config

{'batch_size': 200,
 'checkpoint_path': None,
 'cudnn_deterministic': False,
 'dataset': {'episodes': None,
             'image_transforms': {'enable': False,
                                  'max_num_transforms': 3,
                                  'random_order': False,
                                  'tfs': {'affine': {'kwargs': {'degrees': (-5.0,
                                                                            5.0),
                                                                'translate': (0.05,
                                                                              0.05)},
                                                     'type': 'RandomAffine',
                                                     'weight': 1.0},
                                          'brightness': {'kwargs': {'brightness': (0.8,
                                                                                   1.2)},
                                                         'type': 'ColorJitter',
                                                         'weight': 1.0},
                                          'contrast': {'kwargs': {'contrast': (0.8,
                                                                               1.2)},
                                                       'type': 'ColorJitter',
                                                       'weight': 1.0},
                                          'hue': {'kwargs': {'hue': (-0.05,
                                                                     0.05)},
                                                  'type': 'ColorJitter',
                                                  'weight': 1.0},
                                          'saturation': {'kwargs': {'saturation': (0.5,
                                                                                   1.5)},
                                                         'type': 'ColorJitter',
                                                         'weight': 1.0},
                                          'sharpness': {'kwargs': {'sharpness': (0.5,
                                                                                 1.5)},
                                                        'type': 'SharpnessJitter',
                                                        'weight': 1.0}}},
             'repo_id': ["['small_towel_folding_attempt_3_truncated_384'",
                         "'small_towel_folding_attempt_4_truncated_384'",
                         "'small_towel_folding_attempt_5_truncated_384']"],
             'return_uint8': False,
             'revision': None,
             'root': '/workspace/datasets/smallTowelFolding',
             'streaming': False,
             'use_imagenet_stats': True,
             'video_backend': 'pyav'},
 'env': None,
 'eval': {'batch_size': 50, 'n_episodes': 50, 'use_async_envs': True},
 'eval_freq': 0,
 'job_name': 'multi_task_dit_towel_batch200_80k_h264_v3',
 'log_freq': 200,
 'num_workers': 8,
 'optimizer': {'betas': (0.95, 0.999),
               'eps': 1e-08,
               'grad_clip_norm': 10.0,
               'lr': 2e-05,
               'weight_decay': 0.0},
 'output_dir': PosixPath('/workspace/outputs/multi_task_dit_towel_v1'),
 'peft': None,
 'persistent_workers': True,
 'policy': {'beta_end': 0.02,
            'beta_schedule': 'squaredcos_cap_v2',
            'beta_start': 0.0001,
            'clip_sample': True,
            'clip_sample_range': 1.0,
            'device': 'cuda',
            'do_mask_loss_for_padding': False,
            'drop_n_last_frames': 7,
            'dropout': 0.1,
            'hidden_dim': 512,
            'horizon': 32,
            'image_crop_is_random': True,
            'image_crop_shape': (224, 224),
            'image_resize_shape': (256, 256),
            'input_features': {},
            'integration_method': 'euler',
            'license': None,
            'n_action_steps': 24,
            'n_obs_steps': 2,
            'noise_scheduler_type': 'DDPM',
            'normalization_mapping': {'ACTION': <NormalizationMode.MIN_MAX: 'MIN_MAX'>,
                                      'STATE': <NormalizationMode.MIN_MAX: 'MIN_MAX'>,
                                      'VISUAL': <NormalizationMode.MEAN_STD: 'MEAN_STD'>},
            'num_heads': 8,
            'num_inference_steps': None,
            'num_integration_steps': 100,
            'num_layers': 6,
            'num_train_timesteps': 100,
            'objective': 'diffusion',
            'optimizer_betas': (0.95, 0.999),
            'optimizer_eps': 1e-08,
            'optimizer_lr': 2e-05,
            'optimizer_weight_decay': 0.0,
            'output_features': {},
            'prediction_type': 'epsilon',
            'pretrained_path': None,
            'private': None,
            'push_to_hub': False,
            'repo_id': None,
            'rope_base': 10000.0,
            'scheduler_name': 'cosine',
            'scheduler_warmup_steps': 0,
            'sigma_min': 0.0,
            'tags': None,
            'text_encoder_name': 'openai/clip-vit-base-patch16',
            'timestep_embed_dim': 256,
            'timestep_sampling_alpha': 1.5,
            'timestep_sampling_beta': 1.0,
            'timestep_sampling_s': 0.999,
            'timestep_sampling_strategy': 'beta',
            'tokenizer_max_length': 77,
            'tokenizer_padding': 'max_length',
            'tokenizer_padding_side': 'right',
            'tokenizer_truncation': True,
            'use_amp': False,
            'use_peft': False,
            'use_positional_encoding': False,
            'use_rope': True,
            'use_separate_rgb_encoder_per_camera': False,
            'vision_encoder_lr_multiplier': 0.1,
            'vision_encoder_name': 'openai/clip-vit-base-patch16'},
 'prefetch_factor': 2,
 'rename_map': {},
 'resume': False,
 'reward_model': None,
 'sample_weighting': None,
 'save_checkpoint': True,
 'save_freq': 10000,
 'scheduler': {'name': 'cosine', 'num_warmup_steps': 0},
 'seed': 1000,
 'steps': 80000,
 'tolerance_s': 0.0001,
 'use_policy_training_preset': True,
 'wandb': {'add_tags': True,
           'disable_artifact': False,
           'enable': True,
           'entity': None,
           'mode': None,
           'notes': None,
           'project': 'lerobot',
           'run_id': None}}