GRATH: Gradual Self-Truthifying for Large Language Models.

Weixin Chen, Dawn Song, Bo Li

Browse the full ICML paper archive.