发表机构
Institute of Neuroscience, CAS（中国科学院神经科学研究所）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究通过受控实验审计医学视觉语言模型后训练，发现仅语言模型LoRA SFT在准确率提升有限但损害视觉效用，而扩展适应范围和标准GRPO效果不佳，证据目标在训练集有改进但验证集不稳定，揭示了优化活动与有用视觉行为之间的差距。
AI 中文摘要
医学视觉语言模型（VLM）的后训练通常通过答案准确率来评估。我们在一项受控的Qwen2.5-VL-3B研究（基于PMC-VQA数据集）中，考察了准确率变化和训练目标如何与基于图像的决策相关联。我们比较了仅限语言模型的低秩适应（LoRA）监督微调（SFT）、扩展的多模态适应范围、标准的仅答案组相对策略优化（GRPO），以及一种反事实证据目标。在2000个干净测试问题上，语言模型LoRA SFT使正确图像准确率变化了+1.10个百分点（95%配对自助法置信区间：-0.85至+3.05），而视觉效益事件减少了2.40个百分点，图像敏感性下降了5.60个百分点。配对记录显示，有155个获得和203个丢失的视觉效益事件。更广泛的适应范围产生的正确图像准确率低于语言模型LoRA SFT。标准GRPO产生了混合奖励组和参数更新，其干净测试准确率变化不确定。生成审计揭示，规范选项分数可能遵循与生成答案不同的token路径。当沿着贪婪生成路径取分数时，证据目标在训练集上有所改进；在匹配的训练剂量下，其在验证数据上相对于标准GRPO的优势仍不一致。样本级分析追踪了后训练期间证据分数、决策边界和生成答案的变化。这项实证和测量审计识别了优化活动、目标获取和有用的保留视觉行为之间的差距。
英文摘要
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.
