IDEA-Research/ChatRex-7B

1{2  "architectures": [3    "ChatRexAuxForConditionalGeneration"4  ],5  "auto_map": {6    "AutoConfig": "modeling_chatrex.ChatRexAuxConfig",7    "AutoModelForCausalLM": "modeling_chatrex.ChatRexAuxForConditionalGeneration"8  },9  "ignore_index": -100,10  "image_token_index": 32000,11  "model_type": "chatrex",12  "projector_depth": 2,13  "projector_hidden_act": "gelu",14  "text_config": {15    "_name_or_path": "huggingface_checkpoints/lmsys/vicuna-7b-v1.5",16    "architectures": [17      "LlamaForCausalLM"18    ],19    "max_position_embeddings": 4096,20    "model_type": "llama",21    "pad_token_id": 0,22    "rms_norm_eps": 1e-05,23    "torch_dtype": "bfloat16",24    "vocab_size": 3210425  },26  "torch_dtype": "float32",27  "transformers_version": "4.44.2",28  "vision_aux_config": {29    "optimize_vision_tower_aux": false,30    "type": "OpenCLIPVisionTower",31    "use_last_feat": true,32    "vision_tower": "openclip-convnext-large-d-320-laion2B-s29B-b131K-ft-soup"33  },34  "vision_config": {35    "_name_or_path": "huggingface_checkpoints/openai/clip-vit-large-patch14-336",36    "dropout": 0.0,37    "hidden_size": 1024,38    "image_size": 336,39    "intermediate_size": 4096,40    "model_type": "clip_vision_model",41    "num_attention_heads": 16,42    "num_hidden_layers": 24,43    "patch_size": 14,44    "projection_dim": 76845  },46  "vision_feature_layer": -2,47  "vision_feature_select_strategy": "default",48  "visual_prompt_encoder_config": {49    "add_pos_embedding": true,50    "channel_per_level": [51      192,52      384,53      768,54      153655    ],56    "output_size": 7,57    "pos_embedding_dim": 2880,58    "spatail_scale": 0.25,59    "type": "MultiLevelROIVisualPrompt",60    "with_additional_projection": false61  },62  "visual_prompt_hidden_size": 288063}64