GoodPoint trains LLMs on a new dataset of ICLR papers annotated via author responses to produce valid and actionable feedback, yielding an 83.7% improvement in predicted success rate over base models and outperforming similar-sized LLMs on human feedback matching.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
GoodPoint: Learning Constructive Scientific Paper Feedback from Author Responses
GoodPoint trains LLMs on a new dataset of ICLR papers annotated via author responses to produce valid and actionable feedback, yielding an 83.7% improvement in predicted success rate over base models and outperforming similar-sized LLMs on human feedback matching.