fixed
Large
Model fixes
In January some users reported GPT-J as generating irregularly poor output. After a series of model changes and bug-fixes, we believe we have restored output quality.
Change made in January, published to change-log in February