AI Paper Atlas

Research map / Federated learning and privacy / Personalized and heterogeneous FL

Mitigating Non-IID Drift in Zeroth-Order Federated LLM Fine-Tuning with Transferable Sparsity

Yide Ran, Wentao Guo, Jingwei Sun, Yanzhou Pan, Xiaodong Yu, Hao Wang, Jianwen Xie, Yiran Chen, Denghui Zhang, Zhaozhuo Xu

ICLR 2026 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Federated prompt and LoRA tuning · 1 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
use Meerkat 0.1% sparse ZO instead of full-parameter ZOFull-FedZO · Non-IID avg accuracy, local step 10, LLaMA-3.2-1Baverage test accuracy: 0.699 -> 0.759Improved
use Meerkat 0.1% sparse ZO instead of full-parameter ZOFull-FedZO · Non-IID avg accuracy, local step 10, Qwen2-1.5baverage test accuracy: 0.761 -> 0.805Improved
use Meerkat 0.1% sparse ZO instead of full-parameter ZOFull-FedZO · Non-IID avg accuracy, local step 10, Gemma2-2baverage test accuracy: 0.732 -> 0.772Improved
use Meerkat instead of Lora-FedZO at 100 local stepsLora-FedZO · Non-IID avg accuracy, local step 100, LLaMA-3.2-1Baverage test accuracy: 0.628 -> 0.726Improved
use Meerkat instead of Weight Magnitude sparsity on Gemma2-2b at 10 local stepsWeight Magnitude · Non-IID avg accuracy, local step 10, Gemma2-2baverage test accuracy: 0.779 -> 0.772Hurt
use Meerkat instead of Weight Magnitude sparsity on Gemma2-2b at 30 local stepsWeight Magnitude · Non-IID avg accuracy, local step 30, Gemma2-2baverage test accuracy: 0.770 -> 0.756Hurt
increase local steps from 10 to 100 in Meerkat (ablation)Meerkat · Non-IID avg accuracy, LLaMA-3.2-1Baverage test accuracy: 0.759 -> 0.726Hurt
reduce local step to 1 (high-frequency sync) in Meerkat (ablation)Meerkat · Qwen2-1.5b, Dirichlet 0.5 vs IIDaverage test accuracy over seven tasksImproved
add VPCS early stopping of extreme Non-IID clients (Meerkat-vp) (ablation)Meerkat · Non-IID Dirichlet 0.5, avg over seven tasksaverage test accuracyImproved
random client selection for early stopping instead of GradIP-based selection (ablation)Meerkat-vp · Non-IID Dirichlet 0.5, avg over seven tasksaverage test accuracyHurt
extreme sparsity 1e-3 and 1e-4 in Meerkat at high-frequency communication (ablation)Meerkat · T=1 communicationaccuracyNo 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