VideoChat-TPO / config.json
ynhe
init
16dc4f2
raw
history blame
1.5 kB
{
"_name_or_path": "",
"architectures": [
"MultiModalLLM_PT"
],
"auto_map": {
"AutoConfig": "model_config.VideoChatEConfig",
"AutoModel": "modeling_videochate.MultiModalLLM_PT"
},
"model_config": {
"bridge": {
"extra_num_query_token": 64,
"name": "qformer",
"num_query_token": 32,
"qformer_attention_probs_dropout_prob": 0.1,
"qformer_drop_path_rate": 0.2,
"qformer_hidden_dropout_prob": 0.1
},
"freeze_bridge": false,
"freeze_llm": false,
"freeze_vision_encoder": false,
"llm": {
"lora_alpha": 32,
"lora_dropout": 0.1,
"lora_r": 16,
"name": "mistral_7b",
"pretrained_llm_path": "mistralai/Mistral-7B-Instruct-v0.3",
"use_lora": true,
"hidden_size": 4096
},
"loss": {
"use_vision_regression_loss": false
},
"pretrained_paths": {},
"vision_encoder": {
"name":"vit_l14",
"img_size":224,
"patch_size":16,
"d_model":1024,
"encoder_embed_dim":1024,
"encoder_depth":24,
"encoder_num_heads":16,
"drop_path_rate": 0.0,
"num_frames":16,
"tubelet_size":1,
"use_checkpoint":false,
"checkpoint_num":0,
"return_index":-2,
"vit_add_ln":true,
"pretrained": null
}
},
"torch_dtype": "float32",
"transformers_version": "4.38.0",
"use_flash_attention": true,
"use_cache": true,
"build_decoder":true,
"hidden_size": 4096
}