{
  "model_type": "encoder-decoder",
  "vocab_size": 50257,
  "max_position_embeddings": 1024,
  "encoder_layers": 24,
  "encoder_ffn_dim": 4096,
  "encoder_attention_heads": 16,
  "decoder_layers": 24,
  "decoder_ffn_dim": 4096,
  "decoder_attention_heads": 16,
  "dropout": 0.2,
  "activation_function": "gelu",
  "initializer_range": 0.02
}