vision_encoder/config.json

{
  "_name_or_path": "E:/diffusers/cache/vit",
  "architectures": [
    "Blip2VisionModel"
  ],
  "attention_dropout": 0.0,
  "hidden_act": "quick_gelu",
  "hidden_size": 1024,
  "image_size": 224,
  "initializer_range": 1e-10,
  "intermediate_size": 4096,
  "layer_norm_eps": 1e-05,
  "model_type": "blip_2_vision_model",
  "num_attention_heads": 16,
  "num_hidden_layers": 23,
  "patch_size": 14,
  "qkv_bias": true,
  "torch_dtype": "float32",
  "transformers_version": "4.31.0"
}