Patrick Kenny comments

Results 3 comments of


                                            Patrick Kenny

I'm gonna try taking a look at this one, will update if I get stuck(this is my first issue in this project).

I released the PR anyway, if anything it might be a decent temporary fix.

Also seeing this issue with Qwen based models, https://huggingface.co/dunzhang/stella_en_400M_v5 and the 1.5B variant both have this problem.