search

NolanoOrg / cformers

SoTA Transformers with C-backend for fast inference on your CPU.
MIT License
311 stars 29 forks source link

Generation quality is so low compared to native OA model. #24

Closed HCBlackFox closed 1 year ago

HCBlackFox commented 1 year ago

I think something is wrong with the quantisation script?

Ayushk4 commented 1 year ago

I will have a look at it today.

HCBlackFox commented 1 year ago

I will have a look at it today.

Thank you

Ayushk4 commented 1 year ago

This should have been fixed in #26

HCBlackFox commented 1 year ago

This should have been fixed in #26

Yes, now works very well!

Ayushk4 commented 1 year ago

Thanks.