Mechanistically Guided LoRA Improves Paraphrase Consistency in Medical Vision-Language Models
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866908856908513280 |
|---|---|
| author | Sadanandan, Binesh Behzadan, Vahid |
| author_facet | Sadanandan, Binesh Behzadan, Vahid |
| contents | Medical Vision-Language Models can give different yes or no answers to rephrasings of the same clinical question. We study this in MedGemma-4B using PSF-Med Sadanandan and Behzadan (2025), which provides paraphrase pairs for systematic consistency evaluation on medical VQA. On MIMIC-CXR binary questions (n = 158), the baseline flip rate is 14.6% and mean margin difference is 1.63 logits. We validate that Gemma Scope 2 Sparse Autoencoders (SAEs) transfer to MedGemma activations, achieving R2 ~= 0.997 on both medical and general text (n = 100 prompts each, p < 0.001 for exceeding a 0.95 threshold). We then fine-tune Low-Rank Adaptation (LoRA) adapters with a combined loss that balances paraphrase consistency with answer accuracy. This combined approach prevents mode collapse that occurs with pure consistency training while reducing flip rate from 14.6% to 4.4% (p = 0.002, two-proportion z-test) and margin difference from 1.63 to 0.33 (79.5% reduction). Accuracy remains stable at 84.2% baseline versus 82.3% after training (-1.9pp, not significant). On PadChest Balanced (n = 250), flip rate drops from 13.6% to 7.8%, mean margin difference drops from 1.08 to 0.35 (67.9% reduction), and accuracy increases from 66.4% to 69.4%. A layer-range ablation shows that early layers reduce margin differences more than mechanistically selected middle layers. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_00148 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Mechanistically Guided LoRA Improves Paraphrase Consistency in Medical Vision-Language Models Sadanandan, Binesh Behzadan, Vahid Computer Vision and Pattern Recognition Medical Vision-Language Models can give different yes or no answers to rephrasings of the same clinical question. We study this in MedGemma-4B using PSF-Med Sadanandan and Behzadan (2025), which provides paraphrase pairs for systematic consistency evaluation on medical VQA. On MIMIC-CXR binary questions (n = 158), the baseline flip rate is 14.6% and mean margin difference is 1.63 logits. We validate that Gemma Scope 2 Sparse Autoencoders (SAEs) transfer to MedGemma activations, achieving R2 ~= 0.997 on both medical and general text (n = 100 prompts each, p < 0.001 for exceeding a 0.95 threshold). We then fine-tune Low-Rank Adaptation (LoRA) adapters with a combined loss that balances paraphrase consistency with answer accuracy. This combined approach prevents mode collapse that occurs with pure consistency training while reducing flip rate from 14.6% to 4.4% (p = 0.002, two-proportion z-test) and margin difference from 1.63 to 0.33 (79.5% reduction). Accuracy remains stable at 84.2% baseline versus 82.3% after training (-1.9pp, not significant). On PadChest Balanced (n = 250), flip rate drops from 13.6% to 7.8%, mean margin difference drops from 1.08 to 0.35 (67.9% reduction), and accuracy increases from 66.4% to 69.4%. A layer-range ablation shows that early layers reduce margin differences more than mechanistically selected middle layers. |
| title | Mechanistically Guided LoRA Improves Paraphrase Consistency in Medical Vision-Language Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.00148 |