config.json

{"model_type":"vision_transformer","vocab_size": 50257, "hidden_size": 768, "num_attention_heads": 12, "max_position_embeddings": 1024, "num_hidden_layers": 12, "attention_probs_dropout_prob": 0.1, "hidden_dropout_prob": 0.1, "intermediate_size": 3072, "initializer_range": 0.02}