4 bits quantization of LLaMA using GPTQ, ported to HIP for use in AMD GPUs.
Geek Repo:Geek Repo
Github PK Tool:Github PK Tool