train_config.json

{
    "dataset": {
        "repo_id": "libero_90_image",
        "root": "./data/datasets/libero_90_image",
        "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
                        ]
                    }
                }
            }
        },
        "revision": null,
        "use_imagenet_stats": true,
        "video_backend": "pyav"
    },
    "env": null,
    "policy": {
        "type": "dit_mt",
        "n_obs_steps": 5,
        "normalization_mapping": {
            "VISUAL": "MEAN_STD",
            "STATE": "MIN_MAX",
            "ACTION": "MIN_MAX"
        },
        "input_features": {
            "observation.images.image": {
                "type": "VISUAL",
                "shape": [
                    3,
                    256,
                    256
                ]
            },
            "observation.images.wrist_image": {
                "type": "VISUAL",
                "shape": [
                    3,
                    256,
                    256
                ]
            },
            "observation.state": {
                "type": "STATE",
                "shape": [
                    8
                ]
            },
            "observation.state.joint": {
                "type": "STATE",
                "shape": [
                    7
                ]
            }
        },
        "output_features": {
            "action": {
                "type": "ACTION",
                "shape": [
                    7
                ]
            }
        },
        "device": "cuda",
        "use_amp": false,
        "horizon": 16,
        "n_action_steps": 8,
        "use_pretrained_dino": true,
        "vision_backbone_size": 18,
        "pretrained_backbone_weights": null,
        "avg_pool": true,
        "vision_backbone_norm_name": "group_norm",
        "vision_backbone_norm_num_groups": 16,
        "use_film": true,
        "tokenizer_max_length": 48,
        "freeze_language_pretrained": true,
        "dim_model": 512,
        "n_heads": 8,
        "dropout": 0.1,
        "dim_feedforward": 2048,
        "feedforward_activation": "gelu",
        "n_encoder_layers": 6,
        "n_decoder_layers": 6,
        "noise_scheduler_type": "DDIM",
        "num_train_timesteps": 100,
        "beta_schedule": "squaredcos_cap_v2",
        "beta_start": 0.0001,
        "beta_end": 0.02,
        "prediction_type": "epsilon",
        "clip_sample": true,
        "set_alpha_to_one": true,
        "steps_offset": 0,
        "num_inference_steps": 10,
        "do_mask_loss_for_padding": false,
        "optimizer_lr": 0.0001,
        "optimizer_betas": [
            0.95,
            0.999
        ],
        "optimizer_eps": 1e-08,
        "optimizer_weight_decay": 1e-06,
        "optimizer_grad_clip_norm": 10.0,
        "scheduler_name": "cosine",
        "scheduler_warmup_steps": 500
    },
    "output_dir": "outputs/dit_mt_libero_90_pretrain",
    "job_name": "dit_mt_libero_90_pretrain",
    "resume": true,
    "seed": 1000,
    "num_workers": 4,
    "batch_size": 32,
    "steps": 200000,
    "eval_freq": 0,
    "log_freq": 500,
    "save_checkpoint": true,
    "save_freq": 100000,
    "use_policy_training_preset": true,
    "optimizer": {
        "type": "adamw",
        "lr": 0.0001,
        "weight_decay": 1e-06,
        "grad_clip_norm": 10.0,
        "betas": [
            0.95,
            0.999
        ],
        "eps": 1e-08
    },
    "scheduler": {
        "type": "diffuser",
        "num_warmup_steps": 500,
        "name": "cosine"
    },
    "eval": {
        "n_episodes": 50,
        "batch_size": 50,
        "use_async_envs": false,
        "max_episodes_rendered": 4
    },
    "wandb": {
        "enable": true,
        "disable_artifact": true,
        "project": "paper",
        "entity": "470620104-technical-university-of-munich",
        "notes": null
    }
}