AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Random Policy Valuation is Enough for LLM Reasoning with Verifiable Rewards

Haoran He, Yuxiao Ye, Qingpeng Cai, Chen Hu, Binxing Jiao, Daxin Jiang, Ling Pan

ICLR 2026 · Reinforcement learning › RL from feedback · approach: Scaling RL with verifiable rewards

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 changedWhereMeasuredEffect
use ROVER instead of best baseline (DAPO)Qwen3-8B-Base · AIME 2024pass@1: 20.8 -> 30.6Improved
use ROVER instead of REINFORCE++Qwen3-8B-Base · HMMT 2025pass@1: 7.1 -> 14.6Improved
use ROVER instead of GRPOQwen3-4B-Base · average pass@1pass@1: 35.3 -> 36.5Improved
use ROVER instead of baselines at large kQwen3-8B-Base · AIME24/AIME25/HMMT25 pass@256pass@256: +16.8 over best baselineImproved
use ROVER instead of GRPOQwen3-8B-Base · AIME24 distinct strategiesdiversity: +6.8%Improved
use ROVER instead of DAPO/GRPO baselines on out-of-distribution taskQwen3-8B-Base · GPQA-diamondpass@1: 46.6 -> 50.2Improved
remove the uniform-policy successor Q term (beta=0) (ablation)ROVER · AIME24/25, HMMT25 (Qwen3-4B-Base)pass@kImproved
scale successor Q term too large (beta=5.0) (ablation)ROVER · AIME24/25, HMMT25 (Qwen3-4B-Base)pass@kHurt
vary beta between 0.2 and 1.0 (ablation)ROVER · AIME24/25, HMMT25 (Qwen3-4B-Base)pass@kNo clear effect
use softmax temperature rho=1 vs much lower or higher (ablation)ROVER · Countdown (Qwen2.5-3B)test scoreImproved
use ROVER instead of GRPO with KL 0.01 on CountdownQwen2.5-3B · Countdowndistinct solutions: 3 -> 17Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy