Research map / Reinforcement learning / RL from feedback
REVA-PO: Stabilizing Reinforcement Learning for Chest X-ray Report Generation
Li Guo, Anas Tahir, Z. Jane Wang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add classifier-guided SFT (Stage 2) after warm-up (ablation) | REVA-PO LVLM · MIMIC-CXR test | CheXpert F1: 0.407 -> 0.474 | Improved |
| add RL stage (Stage 3) after Stages 1 and 2 (ablation) | REVA-PO LVLM · MIMIC-CXR test | BLEU-4: 0.118 -> 0.143 | Improved |
| add warm-up (Stage 1) before Stages 2 and 3 (ablation) | REVA-PO LVLM · MIMIC-CXR test | CheXpert F1: 0.481 -> 0.506 | Improved |
| add Response-Weighted Regularization instead of constant KL (ablation) | REVA-PO · MIMIC-CXR test | CheXpert F1: 0.484 -> 0.506 | Improved |
| add KL regularization (ablation) | REVA-PO · MIMIC-CXR test | CheXpert F1: 0.463 -> 0.506 | Improved |
| add Validation-Anchored Policy Reset (ablation) | REVA-PO · MIMIC-CXR test | CheXpert F1: 0.481 -> 0.506 | Improved |
| use BLEU-4 rather than RadGraph F1 as RL reward (ablation) | REVA-PO · MIMIC-CXR test | CheXpert F1: 0.480 -> 0.506 | Improved |
| use CheXpert F1 as RL reward (ablation) | REVA-PO · MIMIC-CXR test | training stability | Hurt |
| increase group size G from 4 to 6 (ablation) | RER · MIMIC-CXR | final metrics | No clear effect |
| scale base LVLM from 3B to 7B (ablation) | REVA-PO · MIMIC-CXR test | BLEU-4: 0.135 -> 0.143 | Improved |
| scale base LVLM from 3B to 7B (ablation) | REVA-PO · IU-Xray test | BLEU-4: 0.195 -> 0.199 | No clear effect |
| replace constant beta with response-weighted beta (compute cost) | REVA-PO · MIMIC-CXR test | step time (s): 11.80 -> 12.02 | Hurt |
Closest papers by meaning
- BenchX: A Unified Benchmark Framework for Medical Vision-Language Pretraining on Chest X-Rays (NeurIPS 2024 · similarity 0.86)
- ReFINE: A Reward-Based Framework for Interpretable and Nuanced Evaluation of Radiology Report Generation (AAAI 2026 · similarity 0.85)
- CXReasonBench: A Benchmark for Evaluating Structured Diagnostic Reasoning in Chest X-rays (NeurIPS 2025 · similarity 0.85)
- Enhanced Contrastive Learning with Multi-view Longitudinal Data for Chest X-ray Report Generation (CVPR 2025 · similarity 0.85)
- Deep Reinforcement Learning for Early Diagnosis of Lung Cancer (AAAI 2024 · similarity 0.85)
- CXPMRG-Bench: Pre-training and Benchmarking for X-ray Medical Report Generation on CheXpert Plus Dataset (CVPR 2025 · similarity 0.84)
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.
