#crscore
NAACL 25: CRScore can produce valid, fine-grained scores of code review quality that have the greatest alignment with human judgment and release a corpus of 2.6k human-annotated review quality scores. arxiv.org/abs/2409.19801 See Atharva's other great work here: scholar.google.com/citations?us...
CRScore: Grounding Automated Evaluation of Code Review Comments in Code Claims and Smells
The task of automated code review has recently gained a lot of attention from the machine learning community. However, current review comment evaluation metrics rely on comparisons with a human-writte...
arxiv.org
January 25, 2025 at 4:20 PM
transforms these signals into training rewards. We show that CRScore++ improves a weaker student model through a combination of supervised fine-tuning and RL critique from a stronger teacher model, thus enabling generalization to novel programming [4/5 of https://arxiv.org/abs/2506.00296v1]
June 3, 2025 at 6:00 AM
CRScore++, an RL framework that leverages both LLM-based subjective feedback and verifiable signals for training. Extending CRScore, a code review evaluation metric integrating LLMs with verifiers like linters and code smell detectors, CRScore++ [3/5 of https://arxiv.org/abs/2506.00296v1]
June 3, 2025 at 6:00 AM
Manav Nitin Kapadnis, Atharva Naik, Carolyn Rose: CRScore++: Reinforcement Learning with Verifiable Tool and AI Feedback for Code Review https://arxiv.org/abs/2506.00296 https://arxiv.org/pdf/2506.00296 https://arxiv.org/html/2506.00296
June 3, 2025 at 6:00 AM