Labrak Yanis
Labrak Yanis
Hi, Since the HugsVision `VisionClassifierTrainer` is based on the HuggingFace transformers library and especially on their `Trainer`, we cannot tweak the loss function. 
HugsVision isn't really suited for reasearch. But you can directly rewrite the Transformers class [ViTForImageClassification](https://github.com/huggingface/transformers/blob/31ec2cb2badfbdd4c1ac9c6c9b8a74e974984206/src/transformers/models/vit/modeling_vit.py#L789) for the `single_label_classification` task. In my opinion, it's the simplest way to do. Clone the...
https://huggingface.co/blog/fine-tune-segformer
Hello, Can you make a pull request with the implementation of it ? It’s will be awesome!
Added with a string parameter in the last commit [fa59fd2](https://github.com/qanastek/HugsVision/commit/fa59fd231a476172c539a7d708ebbe3b6edea710)
Same for me
Thank you @chita216. I do: ```bash pip install --upgrade pip "setuptools
Hi, Thank you very much for your help. I just edit the main branch of the repository to fix this issue. You can install it from here by using `pip...
Thank you @Metal079 and @raseena-tp for your help, and sorry for my late response, I was very busy during past weeks. I push a new commit where I fix the...
> I am also looking for a way to access it, contacted the author but got no answer. Sorry for the late answer, Mickaël ROUVIER will answer to your email...