search

azpoliak / robust-nli

Repository for code from "On Adversarial Removal of Hypothesis-only Bias in Natural Language Inference" (StarSem 2019) and "Don’t Take the Premise for Granted: Mitigating Artifacts in Natural Language Inference" (ACL 2019)
14 stars 5 forks source link

small fix - nltk.word_tokenize returns a list of strings, not just a … #2

Closed pminervini closed 4 years ago

pminervini commented 4 years ago

…single string -> turn that into a string

pminervini commented 4 years ago

now this pull request incudes a simple fix to convert_sick.py that pushes the accuracy from ~26 to ~50

azpoliak commented 4 years ago

Thanks so much for catching this mistake! I am going to open a new issue for us to re-run the models on SICK and then update the paper accordingly

pminervini commented 4 years ago

no prob, this stuff happens all the time :)