ztxz16 / fastllm

纯c++的全平台llm加速库,支持python调用,chatglm-6B级模型单卡可达10000+token / s,支持glm, llama, moss基座,手机端流畅运行
Apache License 2.0
3.31k stars 339 forks source link

Fixing CMake error #309

Open ydm-amazon opened 1 year ago

ydm-amazon commented 1 year ago

Fixes #286 .

laekov commented 1 year ago

Can we leave it configurable like -DCUDA_ARCH=xx?

fluxlinkage commented 4 months ago

Can we leave it configurable like -DCUDA_ARCH=xx?

Latest version (2024/7/10) already supports -DCUDA_ARCH=xx