AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

GIST: Targeted Data Selection for Instruction Tuning via Coupled Optimization Geometry

Guanghui Min, Tianhao Huang, Ke Wan, Chen Chen

ICML 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Adapter-based parameter-efficient tuning · 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
add GIST subspace-alignment selection instead of LESSLlama3.2-3B · TYDIQAaccuracy: 67.1 -> 69.2Improved
use GIST instead of LESSLlama2-7B · MMLU, TYDIQA, BBHaccuracy: LESS 50.2/56.2/41.5 vs GIST 51.2/55.8/41.8Mixed
use GIST instead of LESSQwen2.5-1.5B · BBHaccuracy: 43.2 -> 44.1Improved
select 5% data with GIST vs full-data fine-tuningLlama3.2-3B · Avg. delta over Baseaccuracy: +3.3 -> +4.5Improved
add GIST selection (average over three benchmarks)Qwen2.5-1.5B · Avg. delta over Baseaccuracy: +1.1 (LESS) -> +1.5Improved
perplexity-based selection instead of randomLlama2-7B · Avg. delta over Baseaccuracy: -8.0Hurt
use only one early checkpoint (epoch 1) instead of 4 (ablation)GIST on Llama3.2-3B · TYDIQAaccuracy: 64.3 -> 69.2Improved
aggregate all four checkpoints (ablation)GIST on Llama3.2-3B · TYDIQAaccuracy: 69.2 -> 67.3Hurt
aggregate epochs 1-2 (ablation)GIST on Llama3.2-3B · TYDIQAaccuracy: 69.2 -> 70.2Improved
use full rank 285 instead of rank 150 subspace (ablation)GIST on MMLU · MMLUaccuracy: 62.9 -> 62.5Hurt
rank 10 subspace (40% explained variance) instead of 150 (ablation)GIST on MMLU · MMLUaccuracy: 62.9 -> 62.1No clear effect
single-epoch GIST vs LESS 4-epoch computeGIST vs LESS · runtime and storageA100 hours: LESS 6.0 h + 48.0 h vs GIST 1.5 h + 12.0 hImproved

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