| { | |
| "adaln_group": true, | |
| "attn_cfg": {}, | |
| "attn_layer_idx": [], | |
| "d_intermediate": 0, | |
| "d_model": 1536, | |
| "fused_add_norm": true, | |
| "mixer_drop": 0.0, | |
| "mlp_drop": 0.0, | |
| "n_layer": 48, | |
| "num_classes": 1000, | |
| "num_groups": 4, | |
| "num_tokens": 256, | |
| "pad_vocab_size_multiple": 8, | |
| "residual_in_fp32": true, | |
| "rms_norm": true, | |
| "ssm_cfg": { | |
| "layer": "Mamba2" | |
| }, | |
| "tie_embeddings": true, | |
| "token_drop": 0.0, | |
| "vocab_size": 16384 | |
| } |