| { | |
| "vocab_size": 12536, | |
| "hidden_size": 1024, | |
| "num_hidden_layers": 24, | |
| "num_attention_heads": 16, | |
| "intermediate_size": 2752, | |
| "max_position_embeddings": 1024, | |
| "pad_token_id": 12531, | |
| "mask_token_id": 12535, | |
| "pe_theta": 10000.0, | |
| "layer_norm_eps": 1e-05, | |
| "initializer_range": 0.02, | |
| "tie_word_embeddings": true, | |
| "embed_dropout": 0.0, | |
| "mlp_dropout": 0.0, | |
| "attn_out_dropout": 0.0, | |
| "attn_probs_dropout": 0.0, | |
| "embed_norm": true, | |
| "skip_first_prenorm": true, | |
| "final_norm": true, | |
| "init_method": "megatron" | |
| } |