mlsanityData to 14 Oct 2026

From Reward Signal to Visual Utility: A Controlled Audit of Medical VLM Post-Training

arXiv 2609.31450 ↗First posted 25 Sep 2026cs.CV cs.AIcited by 0

On the map: Multimodal LLMs / vision-language models → Medical vision-language models

Abstract (arXiv)

Medical vision-language model (VLM) post-training is commonly evaluated through answer accuracy. We examine how changes in accuracy and training objectives relate to image-conditioned decisions in a controlled Qwen2.5-VL-3B study on PMC-VQA. We compare supervised fine-tuning (SFT) with low-rank adaptation (LoRA) restricted to the language model, expanded multimodal adaptation scopes, standard answer-only Group Relative Policy Optimization (GRPO), and a counterfactual evidence objective. On 2,000 clean-test questions, language model LoRA SFT changes correct-image accuracy by +1.10 percentage points (95% paired bootstrap CI:-0.85 to +3.05), while visual-benefit events decrease by 2.40 points and image sensitivity decreases by 5.60 points. Paired records reveal 155 acquired and 203 lost visual-benefit events. Broader adaptation yields lower correct-image accuracy than language-model LoRA SFT. Standard GRPO produces mixed-reward groups and parameter updates, with an uncertain clean test accuracy change. A generation audit reveals that canonical option scores can follow a different token path from generated answers. With scores taken along the greedy generation path, the evidence target improves on the training set; its gains over standard GRPO remain inconsistent on validation data at matched training doses. Sample-level analyses trace how evidence scores, decision margins, and generated answers change during post-training. This empirical and measurement audit identifies gaps between optimization activity, target acquisition, and useful held-out visual behavior.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Medical vision-language models, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 13 of its 17 references that are in the map. First posted on arXiv 25 Sep 2026.

Builds on

13 of 17 references.

Closest ancestors · 1

Same direction, newest first

  1. 2026Do Medical Vision Language Models Actually See? A Counterfactual Grounding Framework and Hard-Negative Contrastive Training for Visually-Reliant Medical VLMs4 Jul

Other references · 12

  1. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  2. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  3. 2020Shortcut Learning in Deep Neural Networkscited by 1,969
  4. 2015Sequence Level Training with Recurrent Neural Networkscited by 576
  5. 2021SLAKE: A Semantically-Labeled Knowledge-Enhanced Dataset for Medical Visual Question Answeringcited by 449
  6. 2023PMC-VQA: Visual Instruction Tuning for Medical Visual Question Answeringcited by 338
  7. 2025Perception-Aware Policy Optimization for Multimodal Reasoningcited by 99
  8. 2024"My Answer is C": First-Token Probabilities Do Not Match Text Answers in Instruction-Tuned Language Modelscited by 69
  9. 2025Ground-R1: Incentivizing Grounded Visual Reasoning via Reinforcement Learningcited by 49
  10. 2025MedVLThinker: Simple Baselines for Multimodal Medical Reasoningcited by 27
  11. 2024An Empirical Study on Parameter-Efficient Fine-Tuning for MultiModal Large Language Modelscited by 22
  12. 2026Beyond Accuracy: Evaluating Visual Grounding In Multimodal Medical Reasoningcited by 7

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Toward Effective Reinforcement Learning Fine-Tuning for Medical VQA in Vision-Language Models
  2. 2026Improving Medical Visual Reinforcement Fine-Tuning via Perception and Reasoning Augmentation
  3. 2026Overconfidence and Calibration in Medical VQA: Empirical Findings and Hallucination-Aware Mitigation
  4. 2024SURE-VQA: Systematic Understanding of Robustness Evaluation in Medical VQA Tasks
  5. 2026When Does RL Help Medical VLMs? Disentangling Vision, SFT, and RL Gains
  6. 2025Adaptive Reinforcement for Open-ended Medical Reasoning via Semantic-Guided Reward Collapse Mitigation
  7. 2026Is There Knowledge Left to Extract? Evidence of Fragility in Medically Fine-Tuned Vision-Language Models
  8. 2025Benchmarking and Mitigating Sycophancy in Medical Vision Language Models