| { | |
| "vocab_size": 49152, | |
| "d_model": 2048, | |
| "n_layers": 16, | |
| "n_heads": 16, | |
| "n_kv_heads": 4, | |
| "d_head": 128, | |
| "d_ff": 8192, | |
| "max_seq_len": 8192, | |
| "rope_theta": 10000.0, | |
| "architectures": [ | |
| "CodeTransformer" | |
| ], | |
| "model_type": "code_transformer" | |
| } |
| { | |
| "vocab_size": 49152, | |
| "d_model": 2048, | |
| "n_layers": 16, | |
| "n_heads": 16, | |
| "n_kv_heads": 4, | |
| "d_head": 128, | |
| "d_ff": 8192, | |
| "max_seq_len": 8192, | |
| "rope_theta": 10000.0, | |
| "architectures": [ | |
| "CodeTransformer" | |
| ], | |
| "model_type": "code_transformer" | |
| } |