config.json

{
  "architectures": [
    "LLaMAVIDLlavaForConditionalGeneration"
  ],
  "bert_type": "qformer_pretrain",
  "compress_type": "mean",
  "hidden_size": 4096,
  "ignore_index": -100,
  "image_token_index": 32000,
  "mm_hidden_size": 1408,
  "model_type": "llamavid",
  "num_query": 32,
  "pad_token_id": 32001,
  "projector_hidden_act": "gelu",
  "qformer_config": {
    "model_type": "llamavid_qformer",
    "vocab_size": 30523
  },
  "text_config": {
    "_name_or_path": "C:\\\\Users\\\\niles\\\\OneDrive\\\\Desktop\\\\Data\\\\code\\\\LLaMA-VID\\\\work_dirs\\llama-vid-7b-full-224-video-fps-1",
    "architectures": [
      "LlavaLlamaAttForCausalLM"
    ],
    "bert_type": "qformer_pretrain",
    "compress_type": "mean",
    "freeze_mm_mlp_adapter": false,
    "image_aspect_ratio": "pad",
    "image_grid_pinpoints": null,
    "image_processor": "./llamavid/processor/clip-patch14-224",
    "max_position_embeddings": 4096,
    "max_token": 2048,
    "mm_hidden_size": 1408,
    "mm_projector_type": "mlp2x_gelu",
    "mm_use_im_patch_token": false,
    "mm_use_im_start_end": false,
    "mm_vision_select_feature": "patch",
    "mm_vision_select_layer": -2,
    "mm_vision_tower": "./model_zoo/LAVIS/eva_vit_g.pth",
    "model_type": "llama",
    "num_query": 32,
    "pad_token_id": 0,
    "rms_norm_eps": 1e-05,
    "torch_dtype": "bfloat16",
    "tune_mm_mlp_adapter": false,
    "use_cache": false,
    "use_mm_proj": true,
    "vocab_size": 32064
  },
  "torch_dtype": "bfloat16",
  "transformers_version": "4.40.0.dev0",
  "vision_config": {
    "hidden_size": 1408,
    "image_size": 336,
    "intermediate_size": 4096,
    "model_type": "clip_vision_model",
    "num_attention_heads": 16,
    "num_hidden_layers": 24,
    "patch_size": 14,
    "projection_dim": 768,
    "vocab_size": 32000
  },
  "vision_feature_layer": -2,
  "vision_feature_select_strategy": "default",
  "vision_tower": null,
  "vision_tower_config": {
    "model_type": "llamavid_vision_model"
  },
  "vocab_size": 32064
}