AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

MetaAct-RL: Training Language Models for Reasoning Through Meta-Action-Based Reinforcement Learning

Zhiheng Xi, Yuhui Wang, Yiwen Ding, Guanyu Li, Senjie Jin, Shichun Liu, Jixuan Huang, Dingwen Yang, Jiafu Tang, Boyang Hong, Junjie Ye, Shihan Dou and 7 more

AAAI 2026 · Reinforcement learning › RL from feedback · approach: RL for LLM reasoning · 0 citations (OpenAlex)

Read the paper

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
add length-based reward (ablation)MetaAct-RL (Llama3.2-1B) · AQuAaccuracy: 43.70 -> 48.43Improved
add key-state restart (ablation)MetaAct-RL (Llama3.2-1B) · AQuAaccuracy: 44.49 -> 48.43Improved
restart from forward-reasoning step instead of key critique state (ablation)MetaAct-RL (Llama3.2-1B) · MATHaccuracy: 25.00 -> 23.03Hurt
remove upper bound on length ratio (ablation)length reward (Llama3.2-1B) · MATHaccuracy: 25.00 -> 8.46Hurt
remove lower bound on length ratio (ablation)length reward (Llama3.2-1B) · GSM8Kaccuracy: 52.92 -> 51.25Hurt
MetaAct-RL vs Self-Correction-RLLlama3.2-1B · GSM8Kaccuracy: 46.70 -> 52.92Improved
MetaAct-STaR instead of MetaAct-RL on small modelLlama3.2-1B · 6 benchmark averageaccuracy: 42.23 -> 30.33Hurt
MetaAct-RL vs CoT-RLQwen2.5-1.5B · AIME-2024accuracy: 28.33 -> 35.83Improved
MetaAct-RL vs ADORAQwen2.5-1.5B · AIME-2025accuracy: 25.16 -> 25.42No clear effect

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