AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Towards Efficient Medical Reasoning with Minimal Fine-Tuning Data

Xinlin Zhuang, Feilong Tang, Haolin Yang, Xiwei Liu, Ming Hu, Huifa Li, Haochen Xue, Junjun He, Zongyuan Ge, Yichen Li, Ying Qian, Imran Razzak

CVPR 2026 · Large language models › Instruction and fine-tuning · approach: Fine-tuning for reasoning data · 4 citations (OpenAlex)

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 changedWhereMeasuredEffect
select with difficulty-influence quadrants (DIQ) instead of best baseline selector at 1% budgetLlama3.1-8B-Instruct SFT · HuatuoAvg_A accuracy: 40.54 -> 42.78Improved
select with DIQ instead of best baseline at 10% budgetLlama3.1-8B-Instruct SFT · HuatuoAvg_A accuracy: 41.73 -> 44.04Improved
select with DIQ instead of best baseline at 1% budgetLlama3.1-8B-Instruct SFT · FineMedAvg_A accuracy: 39.20 -> 41.46Improved
select 10% with DIQ instead of PPLLlama3.1-8B-Instruct SFT · FineMedAvg_A accuracy: 38.32 -> 39.36Improved
train on 10% DIQ subset instead of full dataLlama3.1-8B-Instruct SFT · HuatuoAvg_A accuracy: 43.44 -> 44.04Improved
train on 1% DIQ subset instead of full dataLlama3.1-8B-Instruct SFT · FineMedAvg_A accuracy: 38.57 -> 41.46Improved
train on 1% DIQ subset instead of full dataLlama3.1-8B-Instruct SFT · HuatuoAvg_A accuracy: 43.44 vs 42.78, -0.66No clear effect
combine difficulty and influence instead of the strongest single criterion (Reasoning-only) (ablation)DIQ · Huatuo ablation, Llama3.1-8B-InstructAvg_A accuracy: +0.89 at 1%, +0.88 at 10%Improved
enlarge validation set from 90 to 450 examples (ablation)DIQ · Huatuo 1% budgetAvg_A accuracy: 42.04 -> 43.89Improved
reuse Qwen3-8B influence for Qwen3-14B/32B instead of random selectionDIQ · Qwen3-14B, Qwen3-32Baccuracy: +1.42/+1.89/+1.56 on 14B; +0.45/+2.41/+2.64 on 32BImproved
transfer Llama3.1-8B influence to Qwen3 targetsDIQ · Qwen3-8B/14B/32Baccuracy: -0.01 on Qwen3-14B at 1%, -0.41 on Qwen3-32B at 1%, up to +2.11Mixed
apply 1% DIQ SFT before DPO instead of full-data SFTQwen3-8B SFT+DPO · FineMed preference corpusAvg_A accuracy: 55.52 -> 56.52Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy