Research map / Reinforcement learning / RL from feedback
Rate or Fate? RLV$^{\varepsilon}$R: Reinforcement Learning with Verifiable Noisy Rewards
Ali Rad, Khashayar Filom, Darioush Keivan, Peyman Mohajerin Esfahani, Ehsan Kamalinejad
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| verifier with positive Youden index J>0 versus J<0 in GRPO (ablation) | Qwen2.5-3B GRPO on code · OpenR1 code validation | pass@1: J=1.0: +8.0%; J=-0.1: -12.6% vs base | Improved |
| raise verifier quality J from 0.3 to 0.7 to 1.0 (ablation) | Qwen2.5-3B GRPO on code · OpenR1 code validation | pass@1: 16.0% -> 18.6% -> 20.8% (J 0.3, 0.7, 1.0) | Improved |
| use zero-informative verifier J=0 (ablation) | Qwen2.5-3B GRPO on code · OpenR1 code validation | pass@1: +0.6% | No clear effect |
| high false-positive noise instead of high false-negative noise at fixed J=0.3 (ablation) | Qwen2.5-3B GRPO on code · OpenR1 code validation | pass@1: 16.0% -> 14.6% | Hurt |
| anti-aligned verifier J=-0.1 (ablation) | Qwen2.5-3B GRPO on code · OpenR1 code validation | pass@1: 13.4% (J=0) -> 0.16% | Hurt |
| PPO with noisy verifier, J>0 versus J=0 versus J<0 (ablation) | DeepSeek-LLM-7B-chat PPO · GSM8K | validation accuracy: about 24% -> 63-64% for J>0; 21-22% for J~0; 1-3% for J<0 | Improved |
Closest papers by meaning
- Learning to Explore with Parameter-Space Noise: A Deep Dive into Parameter-Space Noise for Reinforcement Learning with Verifiable Rewards (NeurIPS 2026 · similarity 0.92)
- The Reasoning Boundary Paradox: How Reinforcement Learning Constrains Language Models (NeurIPS 2026 · similarity 0.89)
- Not Only Where, But When: Temporal Scheduling for RLVR (NeurIPS 2026 · similarity 0.89)
- What are Key Factors for Updates in RL for LLM Reasoning? (NeurIPS 2026 · similarity 0.89)
- Multi-Step Likelihood-Ratio Correction for Reinforcement Learning with Verifiable Rewards (NeurIPS 2026 · similarity 0.88)
- Exploration vs Exploitation: Rethinking RLVR through Clipping, Entropy, and Spurious Reward (ICLR 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
