Scoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics
Paper • 2609.38847 • Published • 61
Sarah
sbmiller7
·
AI & ML interests
None yet
Recent Activity
upvoted a paper about 14 hours ago
Scoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics updated a model 7 days ago
sbmiller7/vit-classification51 published a model 7 days ago
sbmiller7/vit-classification51Organizations
None yet
upvoted a paper about 14 hours ago