mlsanityData to 7 Oct 2026

Reinforced Attention Learning

arXiv 2602.04884 ↗First posted 4 Feb 2026cs.CL cs.CV cs.LGcited by 5

On the map: LLM reasoning → Multimodal reasoning

Abstract (arXiv)

Post-training with Reinforcement Learning (RL) has substantially improved reasoning in Large Language Models (LLMs) via test-time scaling. However, extending this paradigm to Multimodal LLMs (MLLMs) through verbose rationales yields limited gains for perception and can even degrade performance. We propose Reinforced Attention Learning (RAL), a policy-gradient framework that directly optimizes internal attention distributions rather than output token sequences. By shifting optimization from what to generate to where to attend, RAL promotes effective information allocation and improved grounding in complex multimodal inputs. Experiments across diverse image and video benchmarks show consistent gains over GRPO and other baselines. We further introduce On-Policy Attention Distillation, demonstrating that transferring latent attention behaviors yields stronger cross-modal alignment than standard knowledge distillation. Our results position attention policies as a principled and general alternative for multimodal post-training.

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 33 of its 37 references that are in the map. First posted on arXiv 4 Feb 2026.

Builds on

33 of 37 references.

Closest ancestors · 5

Same direction, newest first

  1. 2025Perception-Aware Policy Optimization for Multimodal Reasoning8 Jul
  2. 2025Visionary-R1: Mitigating Shortcuts in Visual Reasoning with Reinforcement Learning20 May
  3. 2025Video-R1: Reinforcing Video Reasoning in MLLMs27 Mar
  4. 2025Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Models9 Mar
  5. 2022ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoning19 Mar

Other references · 28

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2022Training language models to follow instructions with human feedbackcited by 15,712
  3. 2017Proximal Policy Optimization Algorithmscited by 14,441
  4. 2015Distilling the Knowledge in a Neural Networkcited by 10,076
  5. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
  6. 2023Visual Instruction Tuningcited by 7,792
  7. 2017Deep reinforcement learning from human preferencescited by 3,846
  8. 2023MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Modelscited by 2,689
  9. 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 1,980
  10. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,769
  11. 2023Evaluating Object Hallucination in Large Vision-Language Modelscited by 1,696
  12. 2023MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Modelscited by 1,624
16 more
  1. 2014FitNets: Hints for Thin Deep Netscited by 1,619
  2. 2020MiniLM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformerscited by 1,414
  3. 2024Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysiscited by 1,375
  4. 2023MVBench: A Comprehensive Multi-modal Video Understanding Benchmarkcited by 1,086
  5. 2016Paying More Attention to Attention: Improving the Performance of Convolutional Neural Networks via Attention Transfercited by 1,059
  6. 2018VizWiz Grand Challenge: Answering Visual Questions from Blind Peoplecited by 852
  7. 2023On-Policy Distillation of Language Models: Learning from Self-Generated Mistakescited by 840
  8. 2021NExT-QA:Next Phase of Question-Answering to Explaining Temporal Actionscited by 735
  9. 2024LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understandingcited by 644
  10. 2024BLINK: Multimodal Large Language Models Can See but Not Perceivecited by 529
  11. 2019Patient Knowledge Distillation for BERT Model Compressioncited by 502
  12. 2024MMMU-Pro: A More Robust Multi-discipline Multimodal Understanding Benchmarkcited by 494
  13. 2024TempCompass: Do Video LLMs Really Understand Videos?cited by 352
  14. 2025Video-MMMU: Evaluating Knowledge Acquisition from Multi-Discipline Professional Videoscited by 225
  15. 2024MuirBench: A Comprehensive Benchmark for Robust Multi-image Understandingcited by 178
  16. 2025Diagnosing and Mitigating Modality Interference in Multimodal Large Language Modelscited by 18

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 5.

5 papers.

  1. 2026Beyond Zooming: Learning Multi-Tool Visual Reasoning for Ultra-High-Resolution Remote Sensingcited by 1
  2. 2026Entropy Is Not Enough: Unlocking Effective Reinforcement Learning for Visual Reasoning via Vision-Anchored Token Selectioncited by 0
  3. 2026Information-Regularized Attention for Visual-Centric Reasoningcited by 0
  4. 2026Let Credit Follow Computation: Architecture-Aware Credit Transport for Large Language Model Reinforcement Learningcited by 0
  5. 2026Teach Yourself Where to Look: On-Policy Attention Self-Distillation for Reasoningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Do MLLMs Really See It: Reinforcing Visual Attention in Multimodal LLMs
  2. 2026Faithful-MR1: Faithful Multimodal Reasoning via Anchoring and Reinforcing Visual Attention
  3. 2026Token-level Response-visual Attention Guidance for Multimodal LLMs Knowledge Distillation
  4. 2025Advancing Multimodal Reasoning: From Optimized Cold Start to Staged Reinforcement Learning
  5. 2025Observe-R1: Unlocking Reasoning Abilities of MLLMs with Dynamic Progressive Reinforcement Learning
  6. 2026Visually-Guided Policy Optimization for Multimodal Reasoning
  7. 2026Staying VIGILant: Mitigating Visual Laziness via Counterfactual Visual Alignment in MLLMs
  8. 2026Reinforcing Multimodal Reasoning Against Visual Degradation