{ "datasets": { "assemble_data": { "data_features": "src_pc,tgt_pc,imgs,lang,asset,category,pre_pose", "data_mix": "official_twobytwo_pose_qa+official_twobytwo_pose_instruction_qa", "dataset_py": "hdf5datasets", "default_point_end_token": "", "default_point_patch_token": "", "default_point_start_token": "", "drop_last": true, "eval_with_val": true, "import_rotation": true, "import_translation": false, "label_pad_token_id": -100, "num_samples": 100, "num_workers": 0, "per_device_batch_size": 4, "point_token_len": 256, "shuffle": true, "split_train_val": true, "test_mix": "twobytwo_final_hdf5", "train_mix": "biassembly_final_hdf5+partnet_final_hdf5+partnext_final_hdf5+twobytwo_final_hdf5" }, "point_token_len": 256 }, "framework": { "action_model": { "action_dim": 7, "action_hidden_dim": 1024, "action_horizon": 1, "add_pos_embed": true, "diffusion_model_cfg": { "cross_attention_dim": 2048, "dropout": 0.2, "final_dropout": true, "interleave_self_attention": true, "norm_type": "ada_norm", "num_layers": 16, "output_dim": 1024, "positional_embeddings": null }, "expected_len": 20, "fast_tokenizer_num_samples": 100000, "future_action_window_size": 7, "hidden_size": 1024, "max_seq_len": 1024, "model_path": "/gemini/space/users/jingzhi/project/AssemPVLM-main/starVLA/results/train_vndgcnn_jingzhi15/fast_tokenizer_trained", "noise_beta_alpha": 1.5, "noise_beta_beta": 1.0, "noise_s": 0.999, "num_inference_timesteps": 4, "num_target_vision_tokens": 32, "num_timestep_buckets": 1000, "pad_token_id": 2047, "past_action_window_size": 0, "repeated_diffusion_steps": 8, "scale": 900, "state_dim": 7, "tokenizer_vocab_size": 201, "type": "linear", "use_expected_len": false, "vocab_size": 2048 }, "assemlm_py": "assemlm", "framework_py": "PVLA_Fast", "name": "AssemLM", "point_encoder": { "model_path": "/gemini/space/users/jingzhi/project/AssemPVLM-main/assempvlm/models_local/pretrain_vndgcnn/1000-network_A.pth", "pc_feat_dim": 512, "resume_from_checkpoint": false, "type": "vn_dgcnn" }, "point_projector": { "backbone_output_dim": 12, "project_output_dim": 2048, "projection_hidden_dim": [ 1024, 2048 ], "projection_hidden_layer": 2 }, "use_category_in_instruction": true, "vlm": { "base_vlm": "vlm", "build_version": 0, "local_files_only": true, "saved_version": 1 } }, "is_debug": false, "model_type": "assemlm_hf", "run_id": "assemlm21", "run_root_dir": "./results", "seed": 42, "trackers": [ "jsonl", "wandb" ], "trainer": { "enable_gradient_checkpointing": true, "enable_mixed_precision_training": true, "epochs": 1, "eval_end_step": 1000000, "eval_interval": 200, "eval_start_step": 0, "freeze_modules": null, "gen_do_sample": false, "gen_temperature": 1.0, "gradient_accumulation_steps": 1, "gradient_clipping": 1.0, "is_resume": true, "learning_rate": { "action_model": 0.0001, "base": 1e-05, "pvlm_interface": 5e-06 }, "logging_frequency": 10, "lr_scheduler_type": "cosine_with_min_lr", "max_gen_tokens": 50, "max_grad_norm": 1.0, "max_train_steps": 1000000, "num_warmup_steps": 50000, "optimizer": { "betas": [ 0.9, 0.95 ], "eps": 1e-08, "name": "AdamW", "weight_decay": 1e-08 }, "pretrained_checkpoint": "/gemini/space/users/jingzhi/project/AssemPVLM-main/starVLA/results/train_vndgcnn_jingzhi19/checkpoints/latest/pytorch_model/mp_rank_00_model_states.pt", "resume_epoch": null, "resume_from_checkpoint": "/gemini/space/users/jingzhi/project/AssemPVLM-main/starVLA/results/train_vndgcnn_jingzhi19/checkpoints/latest", "resume_step": null, "save_end_step": 1000000, "save_interval": 200, "save_start_step": 0, "scheduler_specific_kwargs": { "min_lr": 5e-07 }, "train_fast_tokenizer": false, "vlm_max_length": 1800, "vlm_padding": true, "warmup_ratio": 0.1, "weight_decay": 0.0 }, "transformers_version": "4.57.0" }