![[python]windows下安装mamba_ssm简单流程](https://img.php.cn/upload/article/001/503/042/175720370453143.jpg)
numpy-1.24.1-cp310-cp310-win_amd64.whl
torch-2.1.1+cu118-cp310-cp310-win_amd64.whl
torchaudio-2.1.1+cu118-cp310-cp310-win_amd64.whl
torchvision-0.16.1+cu118-cp310-cp310-win_amd64.whl
causal_conv1d-1.1.1-cp310-cp310-win_amd64.whl
mamba_ssm-1.1.3-cp310-cp310-win_amd64.whl
triton-2.1.0-cp310-cp310-win_amd64.whl
部分依赖可通过 pip 直接安装，实际只需提前准备好以下三个关键的 whl 文件：
triton-2.1.0-cp310-cp310-win_amd64.whl
causal_conv1d-1.1.1-cp310-cp310-win_amd64.whl
mamba_ssm-1.1.3-cp310-cp310-win_amd64.whl
上述文件可在网络上搜索获取，或访问 gitee.com/firc/pythonlibs_whl_mirror 下载镜像资源。
准备工作完成后，执行如下步骤：
conda create -n py310 python=3.10 -y conda activate py310 pip install numpy==1.24.1 setuptools==68.2.2 pip install triton-2.1.0-cp310-cp310-win_amd64.whl pip install torch==2.1.1+cu118 torchvision==0.16.1+cu118 torchaudio==2.1.1+cu118 -f https://download.pytorch.org/whl/torch_stable.html pip install causal_conv1d-1.1.1-cp310-cp310-win_amd64.whl pip install mamba_ssm-1.1.3-cp310-cp310-win_amd64.whl
完成上述命令后，环境即已安装配置完毕。最后运行以下代码进行测试验证：
import torch
from mamba_ssm import Mamba
batch, length, dim = 2, 64, 16
x = torch.randn(batch, length, dim).to("cuda")
model = Mamba(
d_model=dim,
d_state=16,
d_conv=4,
expand=2
).to("cuda")
y = model(x)
assert y.shape == x.shape
print('success') 
