{ "type": "paravla", "n_obs_steps": 1, "input_features": { "observation.images.image": { "type": "VISUAL", "shape": [ 3, 256, 256 ] }, "observation.images.image2": { "type": "VISUAL", "shape": [ 3, 256, 256 ] }, "observation.state": { "type": "STATE", "shape": [ 8 ] } }, "output_features": { "action": { "type": "ACTION", "shape": [ 7 ] } }, "device": "cuda", "use_amp": true, "use_peft": false, "push_to_hub": false, "repo_id": "GSR/ParaVLA-LIBERO", "private": null, "tags": null, "license": null, "pretrained_path": null, "chunk_size": 16, "n_action_steps": 8, "use_temporal_ensembling": true, "temporal_ensemble_coeff": 0.9, "max_state_dim": 32, "max_action_dim": 32, "normalization_mapping": { "VISUAL": "IDENTITY", "STATE": "MEAN_STD", "ACTION": "MEAN_STD" }, "vision_model": "facebook/dinov2-large", "text_model": "google-t5/t5-large", "image_size": 224, "hidden_size": 1024, "fusion_heads": 16, "fusion_layers": 4, "dropout": 0.1, "num_denoise_steps": 20, "num_expert_layers": 8, "num_expert_heads": 16, "expert_head_dim": 64, "expert_injection_stride": 1, "expert_injection_min_scale": 1.0, "use_coord_injection": false, "freeze_vision_encoder": false, "freeze_text_encoder": true, "freeze_patch_embed": false, "unfreeze_blocks": 4, "use_lora": false, "lora_r": 64, "lora_alpha": 128, "lora_dropout": 0.05, "lora_target_modules": [ "qkv", "proj", "fc1", "fc2", "patch_embed.proj" ], "tokenizer_max_length": 64, "text_dropout_prob": 0.0, "vision_dim": 1536, "use_goal_predictor": false, "use_wrist_camera": true, "guidance_scale": 1.0, "film_scale_init": 1.0, "optimizer_lr": 0.0001, "backbone_lr_fraction": 0.1, "patch_embed_lr_fraction": 0.5, "optimizer_betas": [ 0.9, 0.95 ], "optimizer_eps": 1e-08, "optimizer_weight_decay": 0.0005, "optimizer_grad_clip_norm": 2.0, "scheduler_warmup_steps": 2000, "scheduler_decay_steps": 45000, "scheduler_decay_lr": 1e-06 }