config.json

{
  "_name_or_path": "naver-clova-ix/donut-base",
  "architectures": [
    "VisionEncoderDecoderModel"
  ],
  "decoder": {
    "d_model": 1024,
    "decoder_attention_heads": 16,
    "decoder_ffn_dim": 4096,
    "decoder_layers": 4,
    "dropout": 0.1,
    "max_position_embeddings": 768,
    "model_type": "mbart",
    "torch_dtype": "float32",
    "vocab_size": 57612
  },
  "encoder": {
    "embed_dim": 128,
    "image_size": [
      1280,
      960
    ],
    "hidden_act": "gelu",
    "hidden_size": 1024,
    "layer_norm_eps": 1e-05,
    "model_type": "donut-swin",
    "num_channels": 3,
    "num_heads": [
      4,
      8,
      16,
      32
    ],
    "num_layers": 4,
    "patch_size": 4,
    "qkv_bias": true,
    "window_size": 10
  },
  "input_size": [
    1280,
    960
  ],
  "is_encoder_decoder": true,
  "max_length": 768,
  "model_type": "vision-encoder-decoder",
  "pad_token_id": 1,
  "tie_word_embeddings": false,
  "torch_dtype": "float32",
  "transformers_version": "4.30.1"
}