AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

GradPruner: Gradient-guided Layer Pruning Enabling Efficient Fine-Tuning and Inference for LLMs

Wei Huang, Anda Cheng, Yinggui Wang

ICLR 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Adapter-based parameter-efficient tuning · 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
add sign-aware layer merging after pruning 10 layers (3 merged) (ablation)GradPruner, Llama3.1-8B FFT · 8-dataset avgavg score: 0.741 -> 0.782Improved
add layer merging (2 merged) (ablation)GradPruner, Llama3.1-8B FFT · 8-dataset avgavg score: 0.767 -> 0.786Improved
add layer merging (3 merged) (ablation)GradPruner, Llama3.1-8B LoRA · 8-dataset avgavg score: 0.755 -> 0.786Improved
add layer merging (2 merged) (ablation)GradPruner, Mistral-7B FFT · 8-dataset avgavg score: 0.740 -> 0.770Improved
prune more than ~10 layers with layer pruning only (ablation)GradPruner, Llama3.1-8B · downstream accuracyaccuracyHurt
merge only same-sign parameters instead of weight averaging (ablation)GradPruner · downstream accuracyaccuracyImproved
prune hidden-size kernels instead of layers (ablation)GradPruner · downstream accuracyaccuracyHurt
accumulate gradients over 0.02% of steps instead of 1% (ablation)GradPruner, Llama3.1-8B FFT · HellaSwagaccuracy: 0.939 -> 0.911Hurt
use 1% of WinoGrande training data instead of 100% (ablation)GradPruner, Llama3.1-8B FFT · WinoGrandeaccuracy: 0.861 -> 0.842Hurt
use GradPruner instead of Laco or SAT at 20% sparsityGradPruner, Llama3.1-8B FFT · HellaSwagaccuracy: 0.942 vs 0.936 (Laco), 0.940 (SAT)No clear effect
use GradPruner instead of other pruners at 40% sparsityLlama3.1-8B FFT · 8-dataset avgavg score: 0.734-0.750 -> 0.782Improved
prune 40% of parameters with GradPrunerLlama3.1-8B FFT · 8-dataset avgavg score: 0.784 -> 0.782Hurt

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