optimum-intel-internal-testing/tiny-mamba

1{
2  "architectures": [
3    "MambaForCausalLM"
4  ],
5  "bos_token_id": 0,
6  "conv_kernel": 3,
7  "d_model": 16,
8  "d_state": 32,
9  "eos_token_id": 0,
10  "expand": 2,
11  "hidden_act": "silu",
12  "hidden_size": 64,
13  "initializer_range": 0.1,
14  "intermediate_size": 128,
15  "layer_norm_epsilon": 1e-05,
16  "model_type": "mamba",
17  "n_layer": 2,
18  "num_hidden_layers": 32,
19  "pad_token_id": 0,
20  "rescale_prenorm_residual": false,
21  "residual_in_fp32": true,
22  "state_size": 16,
23  "time_step_floor": 0.0001,
24  "time_step_init_scheme": "random",
25  "time_step_max": 0.1,
26  "time_step_min": 0.001,
27  "time_step_rank": 4,
28  "time_step_scale": 1.0,
29  "torch_dtype": "float32",
30  "transformers_version": "4.49.0",
31  "use_bias": false,
32  "use_cache": true,
33  "use_conv_bias": true,
34  "use_mambapy": false,
35  "vocab_size": 50280
36}
37