AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

From Exploration to Exploitation: A Two-Stage Entropy RLVR Approach for Noise-Tolerant MLLM Training

Donglai Xu, Hongzheng Yang, Yuzhi Zhao, Pingping Zhang, Jinpeng Chen, Wenao Ma, Zhijian Hou, Mengyang Wu, Xiaolei Li, Senkang Hu, Ziyi Guan, Jason Chun Lok Li and 1 more

CVPR 2026 · Reinforcement learning › RL from feedback · approach: RL for LLM reasoning · 4 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
add two-stage entropy schedule (maximize then minimize) to GRPOGRPO, Qwen2.5-VL-3B · ScreenSpot GUI grounding, 50% label noiseaccuracy (%): 76.2 -> 80.2Improved
add two-stage entropy schedule to GRPOGRPO, Qwen2.5-VL-3B · ScreenSpot GUI grounding, 100% label noiseaccuracy (%): 71.0 -> 75.8Improved
add two-stage entropy schedule to GRPOGRPO, Qwen2.5-VL-3B · ScreenSpot GUI grounding, 0% label noiseaccuracy (%): 82.2 -> 83.6Improved
add constant entropy minimization to GRPO (ablation)GRPO, Qwen2.5-VL-3B · ScreenSpot GUI grounding, 0% label noiseaccuracy (%): 82.2 -> 79.0Hurt
add constant entropy minimization to GRPO (ablation)GRPO, Qwen2.5-VL-3B · ScreenSpot GUI grounding, 100% label noiseaccuracy (%): 71.0 -> 73.2Improved
add constant entropy maximization to GRPO (ablation)GRPO, Qwen2.5-VL-3B · ScreenSpot GUI grounding, 50% label noiseaccuracy (%): 76.2 -> 77.8Improved
add two-stage entropy schedule to GRPOGRPO, Qwen2.5-VL-3B · Pets37 4-shot, 50% label noiseaccuracy (%): 68.5 -> 68.4No clear effect
add two-stage entropy schedule to GRPOGRPO, Qwen2.5-VL-3B · Pets37 4-shot, 100% label noiseaccuracy (%): 54.7 -> 54.3No clear effect
add two-stage entropy schedule to GRPOGRPO, Qwen2-VL-7B · ScreenSpot GUI grounding, 50% label noiseaccuracy (%): 61.2 -> 69.8Improved
add two-stage entropy schedule to GRPOGRPO, Qwen2-VL-2B · ScreenSpot GUI grounding, 50% label noiseaccuracy (%): 32.8 -> 25.2Hurt
add two-stage entropy schedule to GRPOGRPO, Qwen2-VL-2B · OVOD COCO unseen, 50% label noisemAP@0.5: 15.94 -> 19.47Improved
order entropy phases as max then min instead of min then max (ablation)two-stage GRPO, Qwen2.5-VL-3B · ScreenSpot GUI grounding, 100% label noiseaccuracy (%): 70.2 -> 75.8Improved

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