AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

The Appeal and Reality of Recycling LoRAs with Adaptive Merging

Haokun Liu, Gyung Hyun Je, Marco Ciccone, Zhenlin Xu, Prasanth YSS, Colin Raffel

ICML 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: LoRA variants and theory · 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 target-task LoRA to the merging pool (ablation)LoraHub (30) · 62 tasks (Llama 3.1 8B)average accuracy: 0.563 -> 0.663Improved
add target-task LoRA to the merging pool (ablation)Arrow (30) · 62 tasks (Llama 3.1 8B)average accuracy: 0.471 -> 0.613Improved
add target-task LoRA to the merging pool (ablation)pi-Tuning (20) · 62 tasks (Llama 3.1 8B)average accuracy: 0.668 -> 0.668No clear effect
add target-task LoRA to the merging pool (ablation)Ours (30) · 62 tasks (Llama 3.1 8B)average accuracy: 0.652 -> 0.675Improved
use adaptive merging instead of target-task LoRA aloneOurs (30) · 62 tasks (Llama 3.1 8B), no target LoRA in poolaverage accuracy: 0.657 -> 0.652No clear effect
use evaluation-based instead of random LoRA selection (non-adaptive, Simple avg.) (ablation)Simple avg. (30) · 62 tasks, without target-task LoRAaverage accuracy: 0.496 -> 0.543Improved
use evaluation-based instead of random LoRA selection when target-task LoRA is in pool (ablation)adaptive merging · 62 tasksaverage accuracyNo clear effect
replace recycled LoRAs with randomly reinitialized LoRAs (target LoRA kept) (ablation)Ours · 62 tasks, k in 5,10,20,30average accuracyNo clear effect
use gradient-based instead of gradient-free coefficient tuning (ablation)model-level merging · 20 ablation tasks, k=30, no target LoRAaverage accuracy: 0.590 -> 0.607Improved
apply standard regularizers (dropout, weight decay, batch size) to target-task LoRA training (ablation)target-task LoRA · 62 tasksaverage accuracy: 0.672Hurt
use in-house target-task LoRAs instead of recycled hub LoRAs as poolOurs with target LoRA · 62 tasksaverage accuracyImproved
omit top-ranked in-house LoRAs from the pool (m in 5,10,15,20) (ablation)Ours, k=10 · 62 tasksaverage accuracyHurt

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