| { | |
| "_name_or_path": "./checkpoints/cambrian_Qwen2_7B_img", | |
| "architectures": [ | |
| "CambrianQwenForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "audio_input": false, | |
| "bos_token_id": 151643, | |
| "connect_layer": 2, | |
| "connector_depth": 3, | |
| "connector_only": true, | |
| "context_token_num": 16, | |
| "dino_threshold": 0.83, | |
| "drop_threshold": 0.8, | |
| "eos_token_id": 151645, | |
| "frame_pos": false, | |
| "freeze_mm_mlp_adapter": false, | |
| "hidden_act": "silu", | |
| "hidden_size": 3584, | |
| "highres": true, | |
| "highres_connect": false, | |
| "image_aspect_ratio": "pad", | |
| "image_position": 91, | |
| "image_token_len": 144, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 18944, | |
| "is_image_newline": true, | |
| "is_st_sampler": false, | |
| "lowres_token": 8, | |
| "max_num_segments": 24, | |
| "max_position_embeddings": 32768, | |
| "max_window_layers": 28, | |
| "mm_patch_merge_type": "flat", | |
| "mm_projector_lr": null, | |
| "mm_projector_type": "sva", | |
| "mm_use_im_patch_token": false, | |
| "mm_use_im_start_end": false, | |
| "mm_vision_sampler_lr": null, | |
| "mm_vision_select_feature": "patch", | |
| "mm_vision_select_layer": -2, | |
| "mm_vision_tower_aux_list": [ | |
| "siglip/CLIP-ViT-SO400M-14-384", | |
| "facebook/dinov2-giant-res378" | |
| ], | |
| "mm_vision_tower_aux_token_len_list": [ | |
| 576, | |
| 576 | |
| ], | |
| "mm_vision_tower_lr": null, | |
| "model_type": "cambrian_qwen", | |
| "num_attention_heads": 28, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 4, | |
| "num_of_vision_sampler_layers": 10, | |
| "num_query_group": 1, | |
| "pretrained_qformer": "./checkpoints/bert-base-uncased", | |
| "pretraining_tp": 1, | |
| "query_num_list": [ | |
| 144 | |
| ], | |
| "query_type": "Avg_pool", | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": null, | |
| "rope_theta": 1000000.0, | |
| "sliding_window": null, | |
| "spmd_debug": null, | |
| "spmd_fsdp_sharding": null, | |
| "spmd_mesh": null, | |
| "start_of_vision_sampler_layers": 0, | |
| "stride_of_vision_sampler_layers": 3, | |
| "text_input": true, | |
| "tie_word_embeddings": false, | |
| "tokenizer_model_max_length": 8192, | |
| "tokenizer_padding_side": "right", | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.46.0", | |
| "tune_mm_mlp_adapter": false, | |
| "unfreeze_audio_encoder": false, | |
| "unfreeze_mm_compressor": true, | |
| "unfreeze_mm_vision_tower": false, | |
| "use_cache": false, | |
| "use_mm_proj": true, | |
| "use_pos_skipping": false, | |
| "use_sliding_window": false, | |
| "vision_hidden_size": 1024, | |
| "vision_tower_aux_token_len_list": [ | |
| 576, | |
| 576 | |
| ], | |
| "vocab_size": 152064 | |
| } | |