Adibvafa Fallahpour
Adibvafa Fallahpour
Referring to https://github.com/huggingface/transformers/pull/29552, "there's [a test specific to sequence classification that expects all the unfrozen params to be initialized in the range [0.0, 1.0] ](https://app.circleci.com/pipelines/github/huggingface/transformers/87066/workflows/ee804f42-0535-452c-8ee6-38df520712e3/jobs/1130735/parallel-runs/0/steps/0-115) and the initialized values for...
> Could you rebase on main and make sure the CIs are green! 🤗 Of course! It should be good to merge now. There is a failed test for "MobileViTV2ModelTest"...
@ArthurZucker I did some digging on prior decoder model for classification implementations and realized some of them (e.g. gpt2) use caching. It seems the use case is when you want...
@ArthurZucker I merged new main HuggingFace into my branch. Should be ready to merge.
@ArthurZucker I would love a review!
> For finetuning > > > The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/transformers/pr_31155). All of your documentation changes will be reflected on that endpoint. The docs are available until 30...
> hey any updates on this PR? This should merge soon after @ArthurZucker final review.
@amyeroberts I was wondering if you might have any updates on @ArthurZucker? I haven't heard from him in about a month and just wanted to check if everything is alright.
@ArthurZucker Makes sense, I changed the classification head to a linear layer.
@ArthurZucker Pending review!