Research map / Reinforcement learning / RL from feedback
Stabilizing Policy Gradients for Sample-Efficient Reinforcement Learning in LLM Reasoning
Luckeciano Carvalho Melo, Alessandro Abate, Yarin Gal
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 |
|---|---|---|---|
| use aggressive update regime (5x learning rate, 12x smaller batch) with standard GRPO (ablation) | GRPO · MATH / TEST | accuracy | Mixed |
| add curvature-aware token selection (CAPO) to GRPO under aggressive updates | GRPO (A) · MATH | completions needed / stability: 30x fewer completions on MATH | Improved |
| add curvature-aware token selection (CAPO) to GRPO under aggressive updates | GRPO (A) · TEST (8 benchmarks avg) | completions needed / stability: 9x fewer completions | Improved |
| add curvature-aware selection to Dr.GRPO (Dr.CAPO) (ablation) | Dr.GRPO · MATH | accuracy / collapse | Improved |
| add curvature-aware selection to REINFORCE (ReinCAPO) (ablation) | REINFORCE · MATH | accuracy / collapse | Improved |
Closest papers by meaning
- STAPO: Stabilizing Reinforcement Learning for LLMs by Silencing Rare Spurious Tokens (NeurIPS 2026 · similarity 0.91)
- Accelerating RL for LLM Reasoning with Optimal Advantage Regression (NeurIPS 2025 · similarity 0.91)
- SetPO: Set-Level Policy Optimization for Diversity-Preserving LLM Reasoning (ICML 2026 · similarity 0.91)
- One-Way Policy Optimization for Self-Evolving LLMs (ICML 2026 · similarity 0.90)
- Trajectory Bellman Residual Minimization: A Simple Value-Based Method for LLM Reasoning (NeurIPS 2025 · similarity 0.90)
- Stabilizing Policy Optimization via Logits Convexity (NeurIPS 2026 · similarity 0.90)
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.
