valhalla/bart-large-finetuned-squadv1

BART is a seq2seq model intended for both NLG and NLU tasks . BART-large achieves 88.8 EM and 94.6 F1.6 EM . BART can handle sequences with upto 1024 tokens . The model was trained on google colab v100 GPU and fine-tuned on SQuADv1 dataset . The results are actually slightly worse than given in the paper, but BART can be used for question answering tasks . It can be trained on a Google colab with a v100 v100 CPU . BART was propsed in a paper BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehens,