config.json

{
  "_name_or_path": "omni-research/Tarsier-7b",
  "add_transformer_projector": false,
  "architectures": [
    "TarsierForConditionalGeneration"
  ],
  "ignore_index": -100,
  "image_new_idx": 32003,
  "image_newline_idx": 32002,
  "image_token_index": 32000,
  "model_type": "llava",
  "projector_hidden_act": "gelu",
  "text_config": {
    "_name_or_path": "lmsys/vicuna-7b-v1.5",
    "architectures": [
      "LlamaForCausalLM"
    ],
    "max_position_embeddings": 4096,
    "model_type": "llama",
    "pad_token_id": 0,
    "rms_norm_eps": 1e-05,
    "torch_dtype": "float16",
    "use_bpex_rmsnorm": true,
    "use_bpex_rotary": true,
    "use_cache": false,
    "use_ft_flash_attn": true,
    "vocab_size": 32064
  },
  "torch_dtype": "bfloat16",
  "transformers_version": "4.37.2",
  "vision_config": {
    "hidden_size": 1024,
    "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",
  "vocab_size": 32064
}