🩺 Medical NLP

📷 CVPR2026 · 1 paper notes

📌 Same area in other venues: 🧪 ICML2026 (2) · 💬 ACL2026 (47) · 🔬 ICLR2026 (13) · 🤖 AAAI2026 (11) · 🧠 NeurIPS2025 (14)

Towards Efficient Medical Reasoning with Minimal Fine-Tuning Data

This paper proposes the Difficulty-Influence Quadrant (DIQ) data selection strategy, which jointly considers sample difficulty and gradient influence to enable VLM language backbones to match full-data SFT performance using only 1% of curated data, and to surpass full-data training with just 10%.