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)
As noted in https://github.com/azpoliak/robust-nli/pull/2, the experiments on SICK only included the hypothesis without the premises. In turn we need to re-run these experiments and update the paper accordingly.
As noted in https://github.com/azpoliak/robust-nli/pull/2, the experiments on SICK only included the hypothesis without the premises. In turn we need to re-run these experiments and update the paper accordingly.