Question answering on SQuAD 1.1 dataset by finetuning BERT models. Training is supported both on GPU and on Colab TPU.
Primary LanguageJupyter Notebook