{
  "architectures": [
    "MambaInteger50M"
  ],
  "d_model": 512,
  "n_layer": 16,
  "vocab_size": 16384,
  "ssm_cfg": {
    "d_state": 64,
    "dt_rank": 32,
    "dt_min": 0.001,
    "dt_max": 0.1,
    "use_dyadic_scan": true,
    "scale_bits": 15,
    "n_heads": 16,
    "d_head": 32,
    "use_ssd": true,
    "chunk_size": 64
  },
  "training": {
    "seq_len": 1024,
    "batch_size": 64,
    "gradient_accumulation_steps": 2,
    "total_steps": 200000,
    "learning_rate": 6e-4,
    "decay_lr": 6e-4,
    "weight_decay": 0.05,
    "grad_clip": 50.0,
    "num_workers": 0,
    "log_interval": 500,
    "checkpoint_interval": 10000,
    "use_amp": false,
    "use_compile": false
  },
  "rms_norm_eps": 1e-05,
  "use_bitshift_norm": true,
  "weight_quantization": "bitnet_1.58",
  "model_type": "mamba_integer",
  "model_name": "mamba_integer_50m"
}