| { | |
| "vocab_size": 32000, | |
| "d_model": 1024, | |
| "n_heads": 16, | |
| "n_kv_heads": 4, | |
| "n_layers": 24, | |
| "ff_dim": 2304, | |
| "max_seq_len": 2048, | |
| "rope_local_base": 10000.0, | |
| "rope_global_base": 500000.0, | |
| "dropout": 0.0, | |
| "attn_dropout": 0.0, | |
| "init_std": 0.02 | |
| } |
| { | |
| "vocab_size": 32000, | |
| "d_model": 1024, | |
| "n_heads": 16, | |
| "n_kv_heads": 4, | |
| "n_layers": 24, | |
| "ff_dim": 2304, | |
| "max_seq_len": 2048, | |
| "rope_local_base": 10000.0, | |
| "rope_global_base": 500000.0, | |
| "dropout": 0.0, | |
| "attn_dropout": 0.0, | |
| "init_std": 0.02 | |
| } |