| { | |
| "d_model": 4096, | |
| "d_intermediate": 0, | |
| "n_layer": 56, | |
| "vocab_size": 256000, | |
| "ssm_cfg": { | |
| "layer": [ | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| }, | |
| { | |
| "W4A16": "W4A16QMamba2" | |
| }, | |
| { | |
| "W4A8": "W4A8QMamba2" | |
| } | |
| ], | |
| "ngroups": 8, | |
| "chunk_size": 128 | |
| }, | |
| "attn_layer_idx": [], | |
| "attn_cfg": {}, | |
| "rms_norm": true, | |
| "residual_in_fp32": true, | |
| "fused_add_norm": true, | |
| "pad_vocab_size_multiple": 128, | |
| "tie_embeddings": false, | |
| "norm_cfg": { | |
| "norm": [ | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| }, | |
| { | |
| "W4A16": "RMSNorm" | |
| }, | |
| { | |
| "W4A8": "QRMSNorm" | |
| } | |
| ] | |
| }, | |
| "embedding_cfg": { | |
| "layer": "W4O16Embedding" | |
| }, | |
| "lm_head_cfg": { | |
| "layer": "W4A8B16O16Linear" | |
| }, | |
| "is_hybrid": true | |
| } |