AI Paper Atlas

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

Navigating the Pareto Frontier of Alignment: Spectrum-Adaptive Fine-Tuning for LLMs

Yaoyou Fan, Chao Zhang, Xiaoyu Tan, Chenxing Sun, Yu Yuan, Haoyu Feng, Lu Pan, Ke Zeng, Xunliang Cai

ICML 2026 · Large language models › Instruction and fine-tuning · approach: Efficient LLM fine-tuning systems

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
use SFT instead of base model under 50% label noiseQwen2.5-Math-1.5B · Math avg, 50% noiseavg accuracy: 11.0 -> 5.32Hurt
replace SFT with DFT under 50% label noiseQwen2.5-Math-1.5B · Math avg, 50% noiseavg accuracy: 5.32 -> 24.01Improved
use harmonic interpolation instead of SFT under 50% noiseQwen2.5-Math-1.5B · Math avg, 50% noiseavg accuracy: 5.32 -> 23.41Improved
use geometric instead of linear interpolation under 50% noise (ablation)SAFT · Math avg, 50% noiseavg accuracy: 8.35 -> 20.07Improved
increase Geo-SAFT alpha from 0.5 to 2.0 under 50% noise (ablation)Geo-SAFT · Math avg, 50% noiseavg accuracy: 20.07 -> 24.11Improved
replace SFT with Har-SAFTQwen2.5-Math-7B · Math avg (6 benchmarks)avg accuracy: 20.46 -> 31.37Improved
replace DFT with Har-SAFTQwen2.5-Math-7B · Math avg (6 benchmarks)avg accuracy: 30.20 -> 31.37Improved
replace DFT with Har-SAFTQwen3-14B · Math avg (6 benchmarks)avg accuracy: 31.06 -> 33.50Improved
replace DFT with Har-SAFTQwen2.5-Math-1.5B · Math avg (6 benchmarks)avg accuracy: 26.78 -> 27.48No clear effect
replace DFT with Geo-SAFTQwen3-8B · Math avg (6 benchmarks)avg accuracy: 28.82 -> 29.59Mixed
use geometric instead of harmonic interpolation (ablation)SAFT · Math avg, Qwen2.5-Math-7Bavg accuracy: 31.37 -> 29.99Hurt
replace DFT with Har-SAFTQwen2.5-Math-7B · Out-of-domain avgavg accuracy: 27.28 -> 44.66Improved

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