search

ohmeow / blurr

A library that integrates huggingface transformers with the world of fastai, giving fastai devs everything they need to train, evaluate, and deploy transformer specific models.
https://ohmeow.github.io/blurr
Apache License 2.0
289 stars 34 forks source link

Fine-Tuning the BART Large Model for Text Summarization NOT RUNNING #72

Closed lucawagner1510 closed 2 years ago

lucawagner1510 commented 2 years ago

hello, I tried to run the code written in the article: https://towardsdatascience.com/fine-tuning-the-bart-large-model-for-text-summarization-3c69e4c04582 but I think the code is not updated because it does not find for example HF_Seq2SeqBeforeBatchTrans form and other functions. Do you have any suggestions? Many thanks Luca

ohmeow commented 2 years ago

Yup, that's pre-1.0 code.

Your options are:

  1. Run the code in that article with a previous version of blurr (e.g., pip install ohmeow-blurr==0.0.26)
  2. Use the updated summarization module to modify the article's code to work with 1.0

Lmk how it goes.

lucawagner1510 commented 2 years ago

Thank you very much for your very quick reply: I will do the tests and let you know!

ohmeow commented 2 years ago

Feel free to reopen if necessary :)