step_5000/config.json

{
    "d_model": 768,
    "d_intermediate": 0,
    "n_layer": 16,
    "vocab_size": 10002,
    "ssm_cfg": {
        "d_state": 16,
        "d_conv": 4,
        "expand": 2,
        "layer": "Mamba2"
    },
    "attn_layer_idx": [],
    "attn_cfg": {},
    "rms_norm": true,
    "residual_in_fp32": true,
    "fused_add_norm": true,
    "pad_vocab_size_multiple": 8,
    "tie_embeddings": true
}