/GPTQ-for-LLaMa-ROCm

4 bits quantization of LLaMA using GPTQ, ported to HIP for use in AMD GPUs.

Primary LanguagePythonApache License 2.0Apache-2.0

No issues in this repository yet.