train_config.json

{
    "dataset": {
        "repo_id": "lerobot/aloha_sim_insertion_human",
        "root": null,
        "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": "hybrid_act_diffusion",
        "n_obs_steps": 1,
        "normalization_mapping": {
            "VISUAL": "MEAN_STD",
            "STATE": "MEAN_STD",
            "ACTION": "MEAN_STD"
        },
        "input_features": {
            "observation.images.top": {
                "type": "VISUAL",
                "shape": [
                    3,
                    480,
                    640
                ]
            },
            "observation.state": {
                "type": "STATE",
                "shape": [
                    14
                ]
            }
        },
        "output_features": {
            "action": {
                "type": "ACTION",
                "shape": [
                    14
                ]
            }
        },
        "device": "cuda",
        "use_amp": false,
        "chunk_size": 100,
        "n_action_steps": 100,
        "temporal_ensemble_coeff": null,
        "vision_backbone": "resnet18",
        "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
        "replace_final_stride_with_dilation": false,
        "use_group_norm": true,
        "pre_norm": false,
        "dim_model": 512,
        "n_heads": 8,
        "dim_feedforward": 3200,
        "feedforward_activation": "relu",
        "n_encoder_layers": 4,
        "n_decoder_layers": 1,
        "use_vae": true,
        "latent_dim": 32,
        "n_vae_encoder_layers": 4,
        "noise_scheduler_type": "DDPM",
        "num_train_timesteps": 50,
        "beta_schedule": "squaredcos_cap_v2",
        "beta_start": 0.0001,
        "beta_end": 0.02,
        "prediction_type": "epsilon",
        "clip_sample": true,
        "clip_sample_range": 1.0,
        "diffusion_refine_steps": 3,
        "spatial_softmax_num_keypoints": 32,
        "diffusion_down_dims": [
            256,
            512
        ],
        "diffusion_kernel_size": 3,
        "diffusion_n_groups": 8,
        "diffusion_step_embed_dim": 64,
        "diffusion_use_film_scale_modulation": true,
        "hybrid_mode": "sequential",
        "hybrid_weight": 0.5,
        "keyframe_indices": [
            0,
            49,
            99
        ],
        "diffusion_loss_weight": 1.0,
        "dropout": 0.1,
        "kl_weight": 1.0,
        "smoothness_weight": 0.0,
        "smoothness_type": "velocity",
        "optimizer_lr": 1e-05,
        "optimizer_weight_decay": 0.0001,
        "optimizer_lr_backbone": 1e-05,
        "optimizer_lr_diffusion": 0.0001,
        "scheduler_name": "cosine",
        "scheduler_num_warmup_steps": 500,
        "scheduler_num_training_steps": 30000
    },
    "output_dir": "/root/lerobot/lerobot/common/policies/hybrid/outputs/run2",
    "job_name": "hybrid_act_diffusion_train",
    "resume": false,
    "seed": 1000,
    "num_workers": 8,
    "batch_size": 8,
    "steps": 10000,
    "eval_freq": 10000,
    "log_freq": 400,
    "save_checkpoint": true,
    "save_freq": 10000,
    "use_policy_training_preset": true,
    "optimizer": {
        "type": "adamw",
        "lr": 1e-05,
        "weight_decay": 0.0001,
        "grad_clip_norm": 10.0,
        "betas": [
            0.9,
            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
    },
    "wandb": {
        "enable": false,
        "disable_artifact": false,
        "project": "hybrid_policy_training",
        "entity": null,
        "notes": "Hybrid ACT-Diffusion training run",
        "run_id": null
    }
}