| { | |
| "dim": 1280, | |
| "n_layers": 14, | |
| "n_heads": 10, | |
| "n_kv_heads": 10, | |
| "vocab_size": 60729, | |
| "multiple_of": 256, | |
| "ffn_dim_multiplier": 2.6666666666666665, | |
| "norm_eps": 1e-05, | |
| "max_seq_len": 2048, | |
| "dropout": 0.1, | |
| "device": "cuda" | |
| } |
| { | |
| "dim": 1280, | |
| "n_layers": 14, | |
| "n_heads": 10, | |
| "n_kv_heads": 10, | |
| "vocab_size": 60729, | |
| "multiple_of": 256, | |
| "ffn_dim_multiplier": 2.6666666666666665, | |
| "norm_eps": 1e-05, | |
| "max_seq_len": 2048, | |
| "dropout": 0.1, | |
| "device": "cuda" | |
| } |