lr_find() taking so much time when running readme.md example

[CastHash532]

I tried to run the example from Readme.md to fine tune Bert for multi-label Text Classification task on a subset of the News dataset from Kaggle from: https://www.kaggle.com/rmisra/news-category-dataset

I'm running on Colab gpu and even with freeze_transformer_layers set to True, running lr_find() shows 4h on estimated time

is it normal or am I missing something?

notebook https://github.com/CastHash532/kaggle-automl/blob/main/Kaggle_news_NLP_Transformers_BERT.ipynb