model/visualBERT/config.json.1

{
  "architectures": [
    "VisualBertForPreTraining"
  ],
  "attention_probs_dropout_prob": 0.1,
  "bos_token_id": 0,
  "bypass_transformer": false,
  "eos_token_id": 2,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 768,
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "layer_norm_eps": 1e-12,
  "max_position_embeddings": 512,
  "model_type": "visual_bert",
  "num_attention_heads": 12,
  "num_hidden_layers": 12,
  "pad_token_id": 1,
  "special_visual_initialize": true,
  "transformers_version": "4.7.0.dev0",
  "type_vocab_size": 2,
  "visual_embedding_dim": 2048,
  "vocab_size": 30522
}