CARE: Contrastive Anchor-based Rubric Evolution for Large Language Model Post-Training

Researchers propose CARE (Contrastive Anchor-based Rubric Evolution), a method that uses anchor responses from frontier models to dynamically evolve rubrics during LLM post-training, addressing reward hacking in open-ended tasks. In experiments on WildChecklist-9K with Qwen2.5-7B-Base and Qwen2.5-7B-Instruct, CARE achieves state-of-the-art performance on Arena-Hard-2.0, InfoBench, and FollowBench, and is the only method whose win rate against GPT-4.1 anchor responses improves over 300 training steps, with generalization shown on Llama-3.1-8B-Instruct and Qwen3-8B.

arXiv:2609.00892v1 Announce Type: new Abstract: Rubric-based reinforcement learning decomposes open-ended instructions into prompt-specific, flexible rubrics, making it better suited than reinforcement learning with verifiable rewards for post-training LLMs on open-ended tasks. However, static rubrics are inevitably hacked as the policy evolves, and existing dynamic approaches introduce new problems: undirected rubric extraction, unreliable hack detection, and unbounded rubric proliferation. We propose $\textbf{CARE}$ $\textbf{C}$ontrastive $\textbf{A}$nchor-based $\textbf{R}$ubric $\textbf{E}$volution , which grounds every rubric evolution step in a high-quality anchor response generated by a frontier model conditioned on the prompt and its rubrics. At each training step, CARE contrasts the highest-scoring rollout against the anchor, enabling two complementary mechanisms: an Adaptive branch that reactively repairs reward misspecification; and a Chase branch that proactively converts frontier-level quality gaps into sharper rubrics. Together, the two branches $\textbf{maintain discriminative accuracy in the high-reward region}$---the precise region where reward over-optimization mostly originates. Experiments on WildChecklist-9K with Qwen2.5-7B-Base and Qwen2.5-7B-Instruct show that CARE achieves state-of-the-art performance on Arena-Hard-2.0, InfoBench, and FollowBench, and is the $\textbf{only}$ method whose win rate against GPT-4.1 anchor responses shows sustained improvement throughout 300 training steps; additional results on Llama-3.1-8B-Instruct and Qwen3-8B further indicate that CARE generalizes across model families.