config.json

{
    "d_model": 4096,
    "d_intermediate": 0,
    "n_layer": 64,
    "vocab_size": 32768,
    "ssm_cfg": {
        "layer": "Mamba2",
        "ngroups": 8
    },
    "attn_layer_idx": [],
    "attn_cfg": {},
    "rms_norm": true,
    "residual_in_fp32": true,
    "fused_add_norm": true,
    "pad_vocab_size_multiple": 16,
    "tie_embeddings": false
}