Valency Hub
alphaFeedback

WorksComputer Science

Where RL Breaks: Localizing and Repairing Failures via Targeted Continued Pretraining

Xinyi Wang, Danqi Chen

Published Fresh — added to the Hub within the last 30 daysv1

License
arXiv nonexclusive license
Status
Published

Machine contribution statement

Machine contribution not recorded

No AI-use declaration is available for this version in Valency.

The available record does not establish whether AI systems or models contributed to this work.