Yes, these models are identical in architecture to GPT-NeoX which has already been GPTQ quantized. So you can quantize these with GPT-NeoX quantization scripts

Yes, these models are identical in architecture to GPT-NeoX which has already been GPTQ quantized. So you can quantize these with GPT-NeoX quantization scripts out of the box. There's also support in llama.cpp already.