| { | |
| "d_model": 768, | |
| "d_intermediate": 0, | |
| "n_layer": 10, | |
| "vocab_size": 50280, | |
| "ssm_cfg": {}, | |
| "attn_layer_idx": [ | |
| 1, | |
| 3, | |
| 5, | |
| 6, | |
| 8 | |
| ], | |
| "attn_cfg": { | |
| "num_heads": 8 | |
| }, | |
| "rms_norm": true, | |
| "residual_in_fp32": true, | |
| "fused_add_norm": true, | |
| "pad_vocab_size_multiple": 8, | |
| "tie_embeddings": true | |
| } |