config.json

{
  "activation_function": "gelu",
  "architectures": [
    "DeepShallowModel"
  ],
  "attention_heads": 8,
  "decoder_layers": 1,
  "dropout": 0.1,
  "emb_size": 512,
  "encoder_layers": 12,
  "ffn_hid_dim": 2048,
  "max_position_embeddings": 64,
  "model_type": "transformer",
  "src_vocab_size": 10000,
  "tgt_vocab_size": 10000,
  "torch_dtype": "float32",
  "transformers_version": "4.17.0"
}