| { | |
| "vocab_size": 127, | |
| "dim": 256, | |
| "n_layers": 7, | |
| "n_heads": 4, | |
| "n_kv_heads": 1, | |
| "intermediate_size": 704, | |
| "max_seq_len": 256, | |
| "rope_theta": 10000.0, | |
| "rms_norm_eps": 1e-06, | |
| "dropout": 0.1, | |
| "tie_weights": true | |
| } |
| { | |
| "vocab_size": 127, | |
| "dim": 256, | |
| "n_layers": 7, | |
| "n_heads": 4, | |
| "n_kv_heads": 1, | |
| "intermediate_size": 704, | |
| "max_seq_len": 256, | |
| "rope_theta": 10000.0, | |
| "rms_norm_eps": 1e-06, | |
| "dropout": 0.1, | |
| "tie_weights": true | |
| } |