fracapuano/robot_learning_tutorial_diffusion_example_model
017
1{2 "type": "diffusion",3 "n_obs_steps": 2,4 "input_features": {5 "observation.state": {6 "type": "STATE",7 "shape": [8 69 ]10 },11 "observation.images.up": {12 "type": "VISUAL",13 "shape": [14 3,15 480,16 64017 ]18 },19 "observation.images.side": {20 "type": "VISUAL",21 "shape": [22 3,23 480,24 64025 ]26 }27 },28 "output_features": {29 "action": {30 "type": "ACTION",31 "shape": [32 633 ]34 }35 },36 "device": "mps",37 "use_amp": false,38 "push_to_hub": true,39 "repo_id": null,40 "private": null,41 "tags": null,42 "license": null,43 "pretrained_path": null,44 "horizon": 16,45 "n_action_steps": 8,46 "normalization_mapping": {47 "VISUAL": "MEAN_STD",48 "STATE": "MIN_MAX",49 "ACTION": "MIN_MAX"50 },51 "drop_n_last_frames": 7,52 "vision_backbone": "resnet18",53 "crop_shape": [54 84,55 8456 ],57 "crop_is_random": true,58 "pretrained_backbone_weights": null,59 "use_group_norm": true,60 "spatial_softmax_num_keypoints": 32,61 "use_separate_rgb_encoder_per_camera": false,62 "down_dims": [63 512,64 1024,65 204866 ],67 "kernel_size": 5,68 "n_groups": 8,69 "diffusion_step_embed_dim": 128,70 "use_film_scale_modulation": true,71 "noise_scheduler_type": "DDPM",72 "num_train_timesteps": 100,73 "beta_schedule": "squaredcos_cap_v2",74 "beta_start": 0.0001,75 "beta_end": 0.02,76 "prediction_type": "epsilon",77 "clip_sample": true,78 "clip_sample_range": 1.0,79 "num_inference_steps": null,80 "do_mask_loss_for_padding": false,81 "optimizer_lr": 0.0001,82 "optimizer_betas": [83 0.95,84 0.99985 ],86 "optimizer_eps": 1e-08,87 "optimizer_weight_decay": 1e-06,88 "scheduler_name": "cosine",89 "scheduler_warmup_steps": 50090}