AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

ZeroTuning: Unlocking the Initial Token's Power to Enhance Large Language Models Without Training

Feijiang Han, Xiaodong Yu, Jianheng Tang, Delip Rao, Weihua Du, Lyle Ungar

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Task vectors and steering · 9 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 ZeroTuning to Llama-3.1-8B-InstructLlama-3.1-8B-Instruct · Text classification (7 datasets)accuracy: 59.59 -> 71.44Improved
add ZeroTuning to Qwen-2-7BQwen-2-7B · Text classification (7 datasets)accuracy: 55.10 -> 68.19Improved
add ZeroTuning to DeepSeek-R1-14BDeepSeek-R1-14B · Text classification (7 datasets)accuracy: 67.67 -> 71.87Improved
add ZeroTuning to Llama-3.1-8B-InstructLlama-3.1-8B-Instruct · Multiple-choice QA (7 datasets)accuracy: 58.84 -> 61.48Improved
add ZeroTuning to Qwen-2-7BQwen-2-7B · Multiple-choice QA (7 datasets)accuracy: 63.10 -> 64.84Improved
add ZeroTuning to DeepSeek-R1-14BDeepSeek-R1-14B · Multiple-choice QA (7 datasets)accuracy: 60.05 -> 62.20Improved
add ZeroTuning to Llama-3.1-8B-InstructLlama-3.1-8B-Instruct · MT-Benchscore: 7.804 -> 7.966Improved
add ZeroTuning to Llama-2-13B-ChatLlama-2-13B-Chat · MT-Benchscore: 6.650 -> 6.916Improved
ZeroTuning vs Auto-PASTA on Llama-3.1-8B-Instruct classificationZeroTuning · Text classification (7 datasets)accuracy: 63.73 -> 71.44Improved
use head-specific scaling (Up or Down heads) instead of uniform scaling of all heads (ablation)ZeroTuning · SST-2 and MMLUaccuracyImproved
add hybrid Up+Down head scaling on top of Up or Down alone (ablation)ZeroTuning · SST-2 and MMLUaccuracyNo clear effect
use unsupervised entropy-minimization calibration instead of labeled calibration (ablation)ZeroTuning · 5 datasets (Llama-3.1-8B-Instruct)accuracy: 67.52 -> 67.02No clear effect

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