Rubric Curriculum RL: Exploiting the Generation-Verification Gap in Non-Verifiable Domains
cs.LG, cs.AI
Submitted: 2026-11-04
Updated: 2026-11-04
Code: https://github.com/lechmazur/writing
Terminology
Sources
- Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
- Writing-Zero: Bridge the Gap Between Non-verifiable Tasks and Verifiable Rewards
- EvoLM: Self-Evolving Language Models through Co-Evolved Discriminative Rubrics
- Solving Rubik's Cube with a Robot Hand
- Spontaneous Reward Hacking in Iterative Self-Refinement
- Scaling Laws for Reward Model Overoptimization in Direct Alignment Algorithms
- DR Tulu: Reinforcement Learning with Evolving Rubrics for Deep Research
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- GTPO: Stabilizing Group Relative Policy Optimization via Gradient and Entropy Control
- $V_1$: Unifying Generation and Self-Verification for Parallel Reasoners
- School of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMs
- A Unified Pairwise Framework for RLHF: Bridging Generative Reward Modeling and Policy Optimization
- Qwen3 Technical Report
- ArenaRL: Scaling RL for Open-Ended Agents via Tournament-based Relative Ranking
- One Token to Fool LLM-as-a-Judge
- Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena
- Fine-Tuning Language Models from Human Preferences
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis