harrisonvanderbyl / rwkv-cpp-accelerated

A torchless, c++ rwkv implementation using 8bit quantization, written in cuda/hip/vulkan for maximum compatibility and minimum dependencies
MIT License
303 stars 19 forks source link

Remove redundant merges and vocab from a release folder && Add versio… #8

Closed nenkoru closed 1 year ago

nenkoru commented 1 year ago

…n variable into build.sh