Can we get FP4?
catid opened this issue · comments
FP6 doesn't seem to be a useful size. The best models are 70B that we can run, and only 4 bit models will fit in ~40-48GB VRAM
We will support FP5 soon. Yeah, I will try to also support FP4.
An efficient GPU support for LLM inference with x-bit quantization (e.g. FP6,FP5).
catid opened this issue · comments
FP6 doesn't seem to be a useful size. The best models are 70B that we can run, and only 4 bit models will fit in ~40-48GB VRAM
We will support FP5 soon. Yeah, I will try to also support FP4.