| { | |
| "d_model": 768, | |
| "dropout": 0.3, | |
| "ffn_hidden_dim": 2048, | |
| "max_seq_len": 1024, | |
| "n_heads": 12, | |
| "n_kv_heads": 4, | |
| "n_layers": 24, | |
| "vocab_size": 50432 | |
| } |
| { | |
| "d_model": 768, | |
| "dropout": 0.3, | |
| "ffn_hidden_dim": 2048, | |
| "max_seq_len": 1024, | |
| "n_heads": 12, | |
| "n_kv_heads": 4, | |
| "n_layers": 24, | |
| "vocab_size": 50432 | |
| } |