{ "d_model": 4096, "d_intermediate": 0, "n_layer": 56, "vocab_size": 256000, "ssm_cfg": { "layer": "Mamba2" }, "attn_layer_idx": [], "attn_cfg": {}, "rms_norm": true, "residual_in_fp32": false, "pad_vocab_size_multiple": 16, "tie_embeddings": false }