{ "_moe_implementation": "fused", "action_dim": 55, "action_fp32": false, "action_head_dim": 128, "action_num_attention_heads": 32, "action_num_key_value_heads": 8, "adanorm_time": true, "align_params": { "depth": { "block_future_depth_to_action": false, "detach_future_image_feats": true, "dim_head": 32, "dim_out": 1024, "ff_mult": 1, "input_size": 224, "model_type": "MoRGBD", "moge_path": "assets/pretrained/moge-2-vitb-normal/model.pt", "morgbd_path": "assets/pretrained/lingbot-vla-v2-6b/depth/model.pt", "num_backbone_tokens": 256, "num_heads": 4, "num_layers": 1, "token_size": 16, "use_future_depth": true }, "depth_loss_weight": 0.004, "future_depth_loss_weight": 0.004, "llm": { "dim_out": 2560, "image_input_size": 224, "image_token_size": 8 }, "mode": "query", "num_task_tokens": 8, "use_future_video": true, "video": { "attention_mode": "flex_block_causal", "block_suffix_to_future_video": false, "ckpt_path": "assets/pretrained/lingbot-vla-v2-6b/dino_video/teacher_step_10000.pth", "cls_loss_type": "mse", "cls_loss_weight": 0.2, "cls_pool": "last", "config_path": "assets/pretrained/lingbot-vla-v2-6b/dino_video/config.yaml", "cosine_loss_weight": 0.2, "detach_image_feats": true, "dim_head": 32, "dim_out": 1024, "effective_fps": 1.0, "ff_mult": 1, "future_video_loss_weight": 0.004, "input_size": 256, "log_max_samples": 32, "log_scale": 16, "mse_loss_weight": 1.0, "n_blocks": 1, "num_backbone_tokens": 256, "num_future_frames": 1, "num_heads": 4, "num_layers": 1, "share_future_depth_query": true, "use_cls_loss": false, "use_cosine_loss": true, "use_current_patch_loss": true, "use_current_shared_task_proj": true, "use_mse_loss": true, "use_patch_loss": true, "use_shared_future_task_proj": true, "use_smooth_l1_loss": false, "use_warmup_frame": true }, "visual_dir": "outputs/images", "visual_steps": 5000 }, "architectures": [ "LingbotVlaV2Policy" ], "attention_implementation": "flex_cached", "bias_update_speed": 0.0, "chunk_size": 50, "depth_incremental_training": false, "enable_expert_vision": false, "expert_hidden_size": 768, "expert_intermediate_size": 2752, "expert_vision_path": null, "expert_vision_type": null, "final_norm_adanorm": false, "freeze_vision_encoder": false, "incremental_training": false, "initializer_range": 0.02, "loss_type": "L1_fm", "max_action_dim": 55, "max_state_dim": 55, "model_type": "lingbotvla", "moe_implementation": "fused", "n_action_steps": 50, "n_obs_steps": 1, "norm_qkv": false, "nosplit_gate_liner": false, "num_steps": 10, "post_training": true, "precompute_grid_thw": true, "proj_width": 768, "qwen3vl_use_vision_boundaries": true, "reinit_mismatched_weights": false, "return_image_grid_thw": true, "routed_scaling_factor": 4.0, "router_activation": "sigmoid", "router_z_loss_coeff": 0.0001, "separate_time_proj": false, "sequence_wise_loss_coeff": 0.001, "sequence_wise_mode": "per_sequence", "split_fused_experts_from_decoder_fsdp": false, "split_gate_liner": false, "token_moe_intermediate_size": 512, "token_moe_layers": [ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35 ], "token_num_experts": 32, "token_shared_intermediate_size": 704, "token_top_k": 4, "tokenizer_max_length": 72, "tokenizer_path": "assets/pretrained/Qwen3-VL-4B-Instruct", "train_expert_only": false, "train_state_proj": true, "transformers_version": "4.57.3", "use_cache": false, "use_compile": true, "use_lm_head": false, "use_moe": true, "use_qwen3_chat_template": true, "use_qwen3_fixed_grid_cache": true, "use_shared_expert_gate": false, "vit_attn_implementation": "flash_attention_2", "vlm_causal": true, "vlm_family": "qwen3_vl", "vlm_repo_id": null, "vocab_size": 257152 }