{
  "model_type": "minimamba",
  "_name_or_path": "Mamba_500M",
  "architectures": ["MiniMamba"],
  "dim": 1024,
  "num_layers": 54,
  "num_heads": 32,
  "state_dim": 128,
  "num_groups": 1,
  "conv_size": 4,
  "use_mem_eff_path": true,
  "dt_bias": true,
  "D_has_head_dim": true,
  "learnable_init_states": false,
  "ssm_chunk_size": 256,
  "vocab_size": 200064,
  "ffn_dim_multiplier": 2.0,
  "multiple_of": 256,
  "norm_eps": 1e-5,
  "init_use_depth": false,
  "init_base_std": null,
  "init_std_factor": "disabled",
  "hidden_act": "silu",
  "bias": false,
  "torch_dtype": "bfloat16",
  "seed": 1337,
  "init_args": {
    "dt_max": 0.1,
    "dt_min": 0.001,
    "dt_init_floor": 1e-4,
    "A_init_min": 0.01,
    "A_init_max": 16
  },
  "seq_len": 8192,
  "weight_tying": false,
  "dropout": 0.0,
  "num_epochs": 1,
  "global_bsz": 524288,
  "bsz": 1,
  "warmup_steps": 1907,
  "eval_period": 50,
  "save_period": 500,
  "max_lr": 3.0e-4,
  "min_lr": 3.0e-5,
  "max_norm": 1.0,
  "dilation": 1,
  "fsdp": true,
  "ddp": false,
  "mixed_precision": true,
  "cpu_offload": false,
  "sharding_strategy": "full_shard",
  "state_dict_type": "full",
  "auto_wrap_policy": "partial",
  "backward_prefetch": "backward_pre",
  "forward_prefetch": false,
  "sync_module_states": true,
  "use_orig_params": true,
  "device_id": null,
  "precision": {
    "param": "bfloat16",
    "reduce": "bfloat16",
    "buffer": "bfloat16"
  },
  "fsdp_modules": [
    "MambaBlock"
  ],
  "use_activation_checkpointing": true,
  "use_attn": false,
  "softcap": 50.0,
  "torch_compile": false
}