{
    "hidden_size": 4096,
    "num_hidden_layers": 28,
    "num_attention_heads": 16,
    "intermediate_size": 16384,
    "hidden_act": "gelu",
    "hidden_dropout_prob": 0.1,
    "attention_probs_dropout_prob": 0.1,
    "max_position_embeddings": 2048,
    "type_vocab_size": 2,
    "initializer_range": 0.02,
    "model_type": "gpt_neox"
}