| { | |
| "vocab_size": 50257, | |
| "context_length": 1024, | |
| "emb_dim": 1024, | |
| "n_heads": 16, | |
| "n_layers": 22, | |
| "hidden_dim": 2816, | |
| "head_dim": null, | |
| "qk_norm": true, | |
| "n_kv_groups": 4, | |
| "rope_base": 10000.0, | |
| "dtype": "torch.float32" | |
| } |
| { | |
| "vocab_size": 50257, | |
| "context_length": 1024, | |
| "emb_dim": 1024, | |
| "n_heads": 16, | |
| "n_layers": 22, | |
| "hidden_dim": 2816, | |
| "head_dim": null, | |
| "qk_norm": true, | |
| "n_kv_groups": 4, | |
| "rope_base": 10000.0, | |
| "dtype": "torch.float32" | |
| } |