shmsw25 / FActScore

A package to evaluate factuality of long-form generation. Original implementation of our EMNLP 2023 paper "FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generation"
https://arxiv.org/abs/2305.14251
MIT License
292 stars 43 forks source link

improve processing of Perplexity AI outputs #6

Closed martiansideofthemoon closed 1 year ago

martiansideofthemoon commented 1 year ago

make PPL working (removing citation, and etc, a few things that need special care)

martiansideofthemoon commented 1 year ago

@shmsw25 can we close this too, now that we have a dedicated abstain detection? Or did we want to add something to the current function (https://github.com/shmsw25/FActScore/blob/main/factscore/abstain_detection.py#L31)?

shmsw25 commented 1 year ago

Sounds good to me!