latest/model_config.json

{
  "num_classes": 79,
  "dropout": null,
  "init_args": [
    {
      "attention_probs_dropout_prob": 0.1,
      "bos_token_id": 0,
      "coordinate_size": 128,
      "eos_token_id": 2,
      "fast_qkv": false,
      "gradient_checkpointing": false,
      "has_relative_attention_bias": false,
      "has_spatial_attention_bias": false,
      "has_visual_segment_embedding": true,
      "use_visual_backbone": false,
      "hidden_act": "gelu",
      "hidden_dropout_prob": 0.1,
      "hidden_size": 768,
      "image_feature_pool_shape": [
        7,
        7,
        256
      ],
      "initializer_range": 0.02,
      "intermediate_size": 3072,
      "layer_norm_eps": 1e-05,
      "max_2d_position_embeddings": 1024,
      "max_position_embeddings": 514,
      "max_rel_2d_pos": 256,
      "max_rel_pos": 128,
      "model_type": "layoutlmv2",
      "num_attention_heads": 12,
      "num_hidden_layers": 12,
      "output_past": true,
      "pad_token_id": 1,
      "shape_size": 128,
      "rel_2d_pos_bins": 64,
      "rel_pos_bins": 32,
      "type_vocab_size": 1,
      "vocab_size": 250002,
      "init_class": "LayoutXLMModel"
    }
  ],
  "init_class": "LayoutXLMForTokenClassification"
}