I need help with implementing and training/fine-tuning BERT on a multi-class/multi-label text classification task using the HuggingFace library.
I have 3 datasets for sentiment analysis-like tasks where I've already done EDA, preprocessing, and reached F1 score of 73% with other models.
I just need someone who's experienced with BERT that can implement it correctly to get or exceed 75% F1 score.
I assume that the task won't take long especially if you already have a working code for the model already. Please include how long would the task take you to finish it in your proposal.
If you are busy or don't know the task well, please don't bid and waste my time to get yourself a negative review.