Synthetic QA Corpora Generation with Roundtrip Consistency
This work addresses data scarcity in question answering by creating high-quality synthetic datasets, though it builds incrementally on existing BERT-based methods.
The paper tackles the problem of generating synthetic question answering corpora by combining question generation and answer extraction models with roundtrip consistency filtering, resulting in new state-of-the-art performance on NQ and significant improvements on SQuAD2.
We introduce a novel method of generating synthetic question answering corpora by combining models of question generation and answer extraction, and by filtering the results to ensure roundtrip consistency. By pretraining on the resulting corpora we obtain significant improvements on SQuAD2 and NQ, establishing a new state-of-the-art on the latter. Our synthetic data generation models, for both question generation and answer extraction, can be fully reproduced by finetuning a publicly available BERT model on the extractive subsets of SQuAD2 and NQ. We also describe a more powerful variant that does full sequence-to-sequence pretraining for question generation, obtaining exact match and F1 at less than 0.1% and 0.4% from human performance on SQuAD2.