AI Paper Atlas

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

Data Difficulty and the Generalization–Extrapolation Tradeoff in LLM Fine-Tuning

Siyuan Liu, Tinghong Chen, Xinghan Li, Yifei Wang, Jingzhao Zhang

ICML 2026 · Large language models › Instruction and fine-tuning · approach: Efficient LLM fine-tuning systems · 0 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
remove the longest 10% (hardest) examples from the SFT data (ablation)SFT on OpenR1-Math-94k · Math500, small data sizes (2k-5k)accuracy change vs full SFTImproved
remove the shortest 10% (easiest) examples from the SFT data (ablation)SFT on OpenR1-Math-94k · Math500, large data sizes (20k-50k)accuracy change vs full SFTImproved
remove the longest 10% examples from the SFT data at large data size (ablation)SFT on OpenR1-Math-94k · Math500, large data sizesaccuracy change vs full SFTHurt
use harder (longer CoT) training data at fixed data size (ablation)SFT on OpenMath · avg of Math500, AIME24, Minerva Mathaccuracy gain over baseMixed
increase SFT data size at fixed difficulty (ablation)SFT on OpenMath · avg of Math500, AIME24, Minerva Mathaccuracy gain over baseImproved
use a stronger base model (7B instead of 1.5B) (ablation)SFT on OpenMath · avg of Math500, AIME24, Minerva Mathaccuracy gain over baseImproved
increase data size on iGSM synthetic data (ablation)SFT on iGSM, base Ops[2-8]2k · iGSM test (ops 2-20)SFT improvement over baseImproved
fine-tune on very hard iGSM data with small data budget (ablation)SFT on iGSM, base Ops[2-8]2k · iGSM test (ops 2-20)accuracyHurt
fine-tune on easy iGSM data (ablation)SFT on iGSM, base Ops[2-8]2k · iGSM test (ops 2-20)accuracyMixed
replace SFT with DFT token reweighting when data are too hard for the data size (ablation)SFT on iGSM, base Ops[2-8]2k · iGSM, limited data and hard difficultyimprovement over baseImproved
replace SFT with DFT when data size is large (ablation)SFT on iGSM, base Ops[2-8]2k · iGSM, large data sizesimprovement over baseHurt

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