| { |
| "_name_or_path": "state-spaces/mamba-130m", |
| "architectures": [ |
| "MambaForCausalLM" |
| ], |
| "bos_token_id": 1, |
| "conv_kernel": 4, |
| "d_inner": 1536, |
| "d_model": 768, |
| "eos_token_id": 2, |
| "expand": 2, |
| "fused_add_norm": true, |
| "hidden_act": "silu", |
| "hidden_size": 768, |
| "initializer_range": 0.1, |
| "layer_norm_epsilon": 1e-05, |
| "model_type": "mamba", |
| "n_layer": 24, |
| "num_hidden_layers": 24, |
| "pad_vocab_size_multiple": 8, |
| "residual_in_fp32": true, |
| "rms_norm": true, |
| "ssm_cfg": {}, |
| "state_size": 16, |
| "time_step_rank": 48, |
| "torch_dtype": "float32", |
| "transformers_version": "4.38.0.dev0", |
| "use_bias": false, |
| "use_cache": true, |
| "use_conv_bias": true, |
| "vocab_size": 50280 |
| } |
|
|