Final project in 'Practical topics in Machine Learning' course by Dr. Kfir Bar at Bar-Ilan University.
In this project, I chose to experiment with the Transfer Learning technique by using BERT model and the reading comprehension question answering task, that is I picked up pre-trained BERT language model and tried to fine-tune it, with two different fine-tuning techniques, on the Question Answering task using SQuAD2.0 dataset.
My best model has achieved an F1 score of 74.008 and EM score of 69.434 on the dev set.