mlsanityData to 14 Oct 2026

OraPO: Oracle-educated Reinforcement Learning for Data-efficient and Factual Radiology Report Generation

arXiv 2509.18600 ↗First posted 23 Sep 2025cs.CV cs.AI cs.CLcited by 1

Not on the map. Why?

Abstract (arXiv)

Radiology report generation (RRG) aims to automatically produce clinically faithful reports from chest X-ray images. Prevailing work typically follows a scale-driven paradigm, by multi-stage training over large paired corpora and oversized backbones, making pipelines highly data- and compute-intensive. In this paper, we propose Oracle-educated GRPO (OraPO) with a FactScore-based reward (FactS) to tackle the RRG task under constrained budgets. OraPO enables single-stage, RL-only training by converting failed GRPO explorations on rare or difficult studies into direct preference supervision via a lightweight oracle step. FactS grounds learning in diagnostic evidence by extracting atomic clinical facts and checking entailment against ground-truth labels, yielding dense, interpretable sentence-level rewards. Together, OraPO and FactS create a compact and powerful framework that significantly improves learning efficiency on clinically challenging cases, setting the new SOTA performance on the CheXpert Plus dataset (0.341 in F1) with 2--3 orders of magnitude less training data using a small base VLM on modest hardware.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 86 of its 119 references that are in the map. First posted on arXiv 23 Sep 2025.

Builds on

86 of 119 references.

References · 86

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  3. 2024The Llama 3 Herd of Modelscited by 14,715
  4. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  5. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  6. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  7. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  8. 2025Qwen2.5-VL Technical Reportcited by 5,587
  9. 2023Qwen Technical Reportcited by 3,344
  10. 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,810
  11. 2019CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparisoncited by 1,423
  12. 2025Understanding R1-Zero-Like Training: A Critical Perspectivecited by 1,345
38 more
  1. 2023LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Daycited by 1,305
  2. 2023FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generationcited by 1,282
  3. 2024MiniCPM-V: A GPT-4V Level MLLM on Your Phonecited by 1,079
  4. 2024SimPO: Simple Preference Optimization with a Reference-Free Rewardcited by 963
  5. 2025Group Sequence Policy Optimizationcited by 835
  6. 2024Yi: Open Foundation Models by 01.AIcited by 719
  7. 2024DeepSeek LLM: Scaling Open-Source Language Models with Longtermismcited by 642
  8. 2024ORPO: Monolithic Preference Optimization without Reference Modelcited by 562
  9. 2024InternLM2 Technical Reportcited by 397
  10. 2020Generating Radiology Reports via Memory-driven Transformercited by 368
  11. 2025Rubrics as Rewards: Reinforcement Learning Beyond Verifiable Domainscited by 366
  12. 2025MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attentioncited by 263
  13. 2017On the Automatic Generation of Medical Imaging Reportscited by 216
  14. 2025SFT or RL? An Early Investigation into Training R1-Like Reasoning Large Vision-Language Modelscited by 212
  15. 2022Cross-modal Memory Networks for Radiology Report Generationcited by 185
  16. 2023Orca 2: Teaching Small Language Models How to Reasoncited by 159
  17. 2018Hybrid Retrieval-Generation Reinforced Agent for Medical Image Report Generationcited by 141
  18. 2020Improving Factual Completeness and Consistency of Image-to-Text Radiology Report Generationcited by 141
  19. 2023Interactive and Explainable Region-guided Radiology Report Generationcited by 140
  20. 2024DPO Meets PPO: Reinforced Token Optimization for RLHFcited by 127
  21. 2023Dynamic Graph Enhanced Contrastive Learning for Chest X-ray Report Generationcited by 118
  22. 2023R2GenGPT: Radiology Report Generation with Frozen LLMscited by 114
  23. 2024CheXpert Plus: Augmenting a Large Chest X-ray Dataset with Text Radiology Reports, Patient Demographics and Additional Image Formatscited by 106
  24. 2023METransformer: Radiology Report Generation by Transformer with Multiple Learnable Expert Tokenscited by 101
  25. 2025Learning What Reinforcement Learning Can't: Interleaved Online Fine-Tuning for Hardest Questionscited by 97
  26. 2024CT2Rep: Automated Radiology Report Generation for 3D Medical Imagingcited by 96
  27. 2022Improving Chest X-Ray Report Generation by Leveraging Warm Startingcited by 95
  28. 2023PromptMRG: Diagnosis-Driven Prompts for Medical Report Generationcited by 95
  29. 2025On-Policy RL Meets Off-Policy Experts: Harmonizing Supervised Fine-Tuning and Reinforcement Learning via Dynamic Weightingcited by 91
  30. 2025Sample More to Think Less: Group Filtered Policy Optimization for Concise Reasoningcited by 75
  31. 2021Radiology Report Generation with a Learned Knowledge Base and Multi-modal Alignmentcited by 73
  32. 2023KiUT: Knowledge-injected U-Transformer for Radiology Report Generationcited by 62
  33. 2022Cross-modal Prototype Driven Network for Radiology Report Generationcited by 59
  34. 2025UFT: Unifying Supervised and Reinforcement Fine-Tuningcited by 59
  35. 2024Cal-DPO: Calibrated Direct Preference Optimization for Language Model Alignmentcited by 51
  36. 2024Med-MoE: Mixture of Domain-Specific Experts for Lightweight Medical Vision-Language Modelscited by 46
  37. 2022A Medical Semantic-Assisted Transformer for Radiographic Report Generationcited by 44
  38. 2023ORGAN: Observation-Guided Radiology Report Generation via Tree Reasoningcited by 42

50 most cited of 86 All 86 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Multi-Modal Multi-Agent Reinforcement Learning for Radiology Report Generationcited by 1

Similar, not cited

Close in wording, no citation either way

  1. 2026Enhancing Reinforcement Learning for Radiology Report Generation with Evidence-aware Rewards and Self-correcting Preference Learning
  2. 2026Rethinking the Efficiency and Effectiveness of Reinforcement Learning for Radiology Report Generation
  3. 2025RadHiera: Semantic Hierarchical Reinforcement Learning for Medical Report Generation
  4. 2026CXRMate-2: Structured Multimodal Temporal Embeddings and Tractable Reinforcement Learning for Clinically Acceptable Chest X-ray Radiology Report Generation
  5. 2025EditGRPO: Reinforcement Learning with Post-Rollout Edits for Clinically Accurate Chest X-Ray Report Generation
  6. 2026SDR: Set-Distance Rewards for Radiology Report Generation
  7. 2026HERO: Hierarchical Evidential Reasoning Optimization for Radiology Report Generation via Reason-then-Summarize
  8. 2025Reason Like a Radiologist: Chain-of-Thought and Reinforcement Learning for Verifiable Report Generation