AI Paper Atlas

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

FedP²EFT: Federated Learning to Personalize PEFT for Multilingual LLMs

Royson Lee, Minyoung Kim, Fady Rezk, Rui Li, Stylianos I. Venieris, Timothy Hospedales

AAAI 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
add federated learned Bayesian sparse rank selection (FedP²EFT) over FedL2P learning-rate personalizationFedL2P · MasakhaNEWS tir (seen clients)accuracy (%): 45.3±0.7 -> 63.5±0.3Improved
add FedP²EFT over FedL2PFedL2P · MasakhaNEWS xho (unseen clients)accuracy (%): 64.4±0.9 -> 78.5±1.1Improved
add FedP²EFT over FedL2P on a FedDPA-T base modelFedL2P · XNLI ur (seen clients)accuracy (%): 44.8±4.5 -> 63.7±0.2Improved
add FedP²EFT over FedL2PFedL2P · MasakhaNEWS eng (unseen clients)accuracy (%): 90.6±0.1 -> 90.7±0.1No clear effect
add FedP²EFT over LoRALoRA · Fed-Aya fr (unseen clients)METEOR: 0.55 -> 0.22Hurt
add FedP²EFT over FedL2PFedL2P · Fed-Aya ru (unseen clients)METEOR: 0.25 -> 0.34Improved
use non-FL per-client rank selection (BT-LoRA, AdaLoRA) instead of fixed-rank LoRA (ablation)LoRA · MasakhaNEWS and XNLI (seen clients)accuracy (%): XNLI bg 45.8±0.3 -> 43.8±0.0 (BT-LoRA)Hurt
use FedP²EFT instead of FedL2P for federated training of the personalization strategy generatorFedL2P · XNLI (mBERT) federated trainingmean latency (s): 28.2 -> 3.4Improved
use FedP²EFT instead of FedL2P for federated trainingFedL2P · Fed-Aya (Llama-3 3B) federated trainingmean latency (s): 226.5 -> 80.3Improved
use FedP²EFT instead of FedL2P for federated training peak memoryFedL2P · Fed-Aya (Llama-3 3B) federated trainingpeak memory (GB): 32.2 -> 25.2Improved
use FedP²EFT instead of LoRA at inferenceLoRA · Fed-Aya (Llama-3 3B) inferencepeak memory (GB): 17.9 -> 15.7Improved
use FedP²EFT instead of LoRA at inferenceLoRA · Fed-Aya (Llama-3 3B) inferencemean latency (s): 347.3 -> 400.6Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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