config.json

{
  "arch": "mini_gpt4_llama_v2",
  "architectures": [
    "MiniGPT4_Video"
  ],
  "auto_map": {
    "AutoConfig": "mini_gpt4_llama_v2.minigpt4_video_config",
    "AutoModel": "mini_gpt4_llama_v2.MiniGPT4_Video"
  },
  "chat_template": true,
  "ckpt": "checkpoints/video_llama_checkpoint_last.pth",
  "device": "cuda",
  "drop_path_rate": 0,
  "end_sym": "</s>",
  "freeze_qformer": true,
  "freeze_vit": true,
  "image_size": 224,
  "img_size": 224,
  "length": 50,
  "llama_model": "meta-llama/Llama-2-7b-chat-hf",
  "lora_alpha": 16,
  "lora_dropout": 0.05,
  "lora_r": 64,
  "lora_target_modules": [
    "q_proj",
    "v_proj"
  ],
  "low_resource": true,
  "max_context_len": 3600,
  "max_txt_len": 256,
  "model_type": "minigpt4_video",
  "num_query_token": 32,
  "prompt": "",
  "prompt_path": "",
  "remove_template": false,
  "token_pooling": true,
  "torch_dtype": "float32",
  "transformers_version": "4.42.3",
  "use_grad_checkpoint": true,
  "use_grad_checkpoint_llm": true,
  "vit_model": "eva_clip_g",
  "vit_precision": "fp16"
}