config.json

{
  "activation_dropout": 0.0,
  "activation_function": "gelu",
  "architectures": [
    "MegatronForCausalLM"
  ],
  "attention_dropout": 0.1,
  "bos_token_id": 0,
  "classifier_dropout": 0.0,
  "d_model": 3072,
  "decoder_attention_heads": 32,
  "decoder_ffn_dim": 18432,
  "decoder_layerdrop": 0,
  "decoder_layers": 72,
  "decoder_start_token_id": 0,
  "dropout": 0.1,
  "eos_token_id": 2,
  "gradient_checkpointing": false,
  "init_std": 0.02,
  "is_decoder": true,
  "max_position_embeddings": 1024,
  "model_type": "megatron",
  "pad_token_id": 1,
  "scale_embedding": true,
  "transformers_version": "4.4.0.dev0",
  "use_cache": true,
  "vocab_size": 51200
}