unum-cloud / uform

Pocket-Sized Multimodal AI for content understanding and generation across multilingual texts, images, and 🔜 video, up to 5x faster than OpenAI CLIP and LLaVA 🖼️ & 🖋️
https://unum-cloud.github.io/uform/
Apache License 2.0
1.01k stars 60 forks source link

Can not load multilingual model. (ERROR in huggingface transformers library) #27

Closed javiabellan closed 1 year ago

javiabellan commented 1 year ago

At calling

model = uform.get_model('unum-cloud/uform-vl-multilingual')

which calls at line 293

tokenizer = AutoTokenizer.from_pretrained('sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2')

generates the following error:

fbe2d604-4648-4252-9cc2-2f092b9c6485

NOTE: I have the latest version of transformers lib (4.30.2)

javiabellan commented 1 year ago

I think the file vocab.txt is missing in https://huggingface.co/sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2/tree/main but i dont kown if this is the issue

kimihailv commented 1 year ago

Hello. I tried to load the model and tokenizer and everything was ok (I used the same version of transformers and the latest version of uform from pypi). Could you please reinstall transformers and uform and try again?

ashvardanian commented 1 year ago

@javiabellan, do you still experience this issue?

ashvardanian commented 1 year ago

:tada: This issue has been resolved in version 0.3.1 :tada:

The release is available on GitHub release

Your semantic-release bot :package::rocket: