AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Implicit Actor Critic Coupling via a Supervised Learning Framework for RLVR

Jiaming Li, Longze Chen, Ze Gong, Yukun Chen, Lu Wang, Wanwei He, Run Luo, Minzheng Wang, Lei Zhang, Haoran Ye, Min Yang

ICML 2026 · Reinforcement learning › RL from feedback · approach: Scaling RL with verifiable rewards · 0 citations (OpenAlex)

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 PACS supervised formulation instead of GRPOQwen3-8B · BeyondAIMEpass@64: 57.64 -> 64.02Improved
Use PACS instead of PPOQwen3-8B · BeyondAIMEpass@64: 53.80 -> 64.02Improved
Train Qwen3-4B with PACSQwen3-4B · AIME 2025pass@1: 34.71 -> 45.63Improved
Add class-imbalance weighting to PACS (ablation)PACS Qwen3-8B · BeyondAIMEpass@64: 58.97 -> 64.02Improved
Add class-imbalance weighting to PACS (ablation)PACS Qwen3-4B · AMC 23pass@1: 89.18 -> 90.45Improved
Add class-imbalance weighting to PACS (ablation)PACS Qwen3-8B · AMC 23 and AIME 2024pass@1: AIME 2024: 58.93 (w/o weight) vs 57.58No clear effect
Use RLOO score function instead of GRPO or Dr. GRPO group-mean score (ablation)PACS · BeyondAIMEpass@1: 26.41 / 26.67 -> 27.16Improved
Vary beta in {0.1, 0.5, 1, 2, 10} (ablation)PACS · AMC23, AIME, BeyondAIMEpass@kNo clear effect
Train with PACS instead of GRPO or PPO, evaluate out-of-domainPACS Qwen3-8B · LiveCodeBenchscore: 38.01 -> 45.96Improved
Train with PACS instead of GRPO, evaluate on GPQAPACS Qwen3-8B · GPQAscore: 48.55 -> 52.46Improved
Use PACS instead of GRPOQwen3-8B · AIME 2025 response diversitydiversity score: 0.0249 -> 0.0392Improved
Use PACS instead of GRPO with continuous reward model supervisionQwen3-4B · AIME 2025pass@1: 37.08 -> 40.91Improved

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