config.json

{
  "_attn_implementation_autoset": true,
  "bos_token_id": 0,
  "chunk_size": 256,
  "conv_kernel": 4,
  "eos_token_id": 3,
  "expand": 2,
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_size": 8192,
  "initializer_range": 0.1,
  "is_bitnet_config": true,
  "layer_norm_epsilon": 1e-05,
  "model_type": "mamba2",
  "n_groups": 8,
  "num_heads": 128,
  "num_hidden_layers": 128,
  "pad_token_id": 2,
  "quantization_config": {
    "modules_to_not_convert": [
      "lm_head"
    ],
    "quant_method": "bitnet"
  },
  "rescale_prenorm_residual": false,
  "residual_in_fp32": true,
  "rms_norm": true,
  "state_size": 2048,
  "tie_word_embeddings": false,
  "time_step_floor": 0.0001,
  "time_step_limit": [
    0.0,
    Infinity
  ],
  "time_step_max": 0.1,
  "time_step_min": 0.001,
  "time_step_rank": 256,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.50.3",
  "use_bias": false,
  "use_cache": true,
  "use_conv_bias": true,
  "vocab_size": 100462
}