Could we get a int3 version of gptq please?
#1
by
davidsyoung
- opened
As title. This would be really useful for VRAM constrained workloads. Thank you!
Sorry, you'll need to generate the model yourself, as our uploading process has become significantly more complex for certain reasons.