Thank you. As I'm new to this field, I have been grappling with some concepts like GGML vs GPTQ models, etc. It seems the GPTQ models only run on Linux, so your

Thank you. As I'm new to this field, I have been grappling with some concepts like GGML vs GPTQ models, etc. It seems the GPTQ models only run on Linux, so your suggestion is right to use llama.cpp as that works with GGML and is compatible with M1 Macs as well. I had trouble running oobabooga on M1 Mac, something about Python arm64 vs x86_64... I didn't know about GPT4-x-Vicuna that you mentioned. Interesting to see that it's uncensored. I knew about Wizard-Vicuna-uncensored though. It seems like GPT4-x-Vicuna is similar.