Automatic Curriculum Expert Iteration for Reliable LLM Reasoning

Zirui Zhao, Hanze Dong, Amrita Saha +2

Hallucinations (i.e., generating plausible but inaccurate content) and laziness (i.e. excessive refusals or defaulting to "I don't know") persist as major challenges in LLM reasoni…