View Code? Open in Web Editor
NEW
Code and data for the AAAI 2022 paper "Hybrid Autoregressive Inference for Scalable Multi-hop Explanation Regeneration"
Home Page: https://arxiv.org/abs/2107.11879
hybrid_autoregressive_inference's People
Contributors
Watchers
hybrid_autoregressive_inference's Issues
hello
If the corpus size is 170m, how to modify the process?
bm25_fit, extract_chains.py is very slow
thanks
Hello, I noticed that in your code BM25 is calculated by obtaining a feature vector from the BM25 score of each word in the sentence.
Does calculating the similarity between feature vectors work as well as traditional BM25 scores?
Thanks