config.json

{
  "_name_or_path": "OpenGVLab/InternVL2_5-1B",
  "architectures": [
    "InternVL2_5ForConditionalGeneration"
  ],
  "downsample_ratio": 0.5,
  "dynamic_image_size": true,
  "eos_token_id": 151645,
  "force_image_size": 448,
  "hidden_size": 896,
  "image_end_token_id": 151666,
  "image_start_token_id": 151665,
  "image_token_id": 151667,
  "max_dynamic_patch": 12,
  "min_dynamic_patch": 1,
  "model_type": "internvl2_5",
  "num_image_token": 256,
  "pixel_shuffle_version": "v2",
  "select_layer": -1,
  "text_config": {
    "_attn_implementation_autoset": true,
    "_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
    "architectures": [
      "Qwen2ForCausalLM"
    ],
    "bos_token_id": 151643,
    "eos_token_id": 151645,
    "hidden_size": 896,
    "intermediate_size": 4864,
    "max_window_layers": 21,
    "model_type": "qwen2",
    "num_attention_heads": 14,
    "num_hidden_layers": 24,
    "num_key_value_heads": 2,
    "rope_theta": 1000000.0,
    "torch_dtype": "bfloat16",
    "use_bfloat16": true,
    "vocab_size": 151674
  },
  "tie_word_embeddings": false,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.49.0.dev0",
  "use_thumbnail": true,
  "vision_config": {
    "_attn_implementation_autoset": true,
    "architectures": [
      "InternVL2_5VisionModel"
    ],
    "hidden_size": 1024,
    "image_size": 448,
    "initializer_factor": 1.0,
    "intermediate_size": 4096,
    "model_type": "internvl2_5",
    "norm_type": "layer_norm",
    "num_attention_heads": 16,
    "num_hidden_layers": 24,
    "qk_normalization": false,
    "qkv_bias": true,
    "torch_dtype": "bfloat16"
  }
}