patrick-lewis

3 ms·

1. 1.▲
   by patrick-lewis 6y ago

   Hi, First author here. Thanks for the comment and for adding examples, and for your nuanced comments of the answer overlap split. My position is that these datasets are still useful for QA, but what was lacking was an analysis of how easy&#
2. 2.▲
   by patrick-lewis 6y ago

   Hi, first author here. > Given that nearest-neighbor outperforms on closed book, is it reasonable to suspect the model is doing NN itself internally (which would explain the good performance on close duplicates?) I think this is definite
3. 3.▲
   by patrick-lewis 6y ago

   Hi, first author here. Yes, we evaluated 3 retrieval-based models, DPR, RAG and FID - check out the paper for the numbers ( https://arxiv.org/pdf/2008.02637.pdf )