AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

M-Loss: Quantifying Model Merging Compatibility with Limited Unlabeled Data

Tiantong Wang, Yiyang Duan, Haoyu Chen, Tiantong Wu, Wei Yang Bryan Lim

AAAI 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: Joint pruning and quantization · 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
replace fixed TIES trimming with M-Loss-guided dynamic pruning (M-TIES)TIES · 8-task merging, ViT-B/32average accuracy: 72.42 -> 73.23Improved
replace fixed TIES trimming with M-TIESTIES · 8-task merging, ViT-L/14average accuracy: 84.94 -> 85.28Improved
M-TIES vs ensemblingM-TIES · 8-task merging, ViT-B/32average accuracy: 78.56 -> 73.23Hurt
M-TIES vs ensemblingM-TIES · 8-task merging, ViT-L/14average accuracy: 85.56 -> 85.28Hurt
replace TIES with M-TIES on SVHNTIES · SVHN, ViT-B/32accuracy (%): 86.20 -> 83.06Hurt
M-TIES on out-of-domain datasets vs TIESTIES · OOD average, ViT-B/32accuracy (%): 57.55 -> 63.46Improved
increase unlabeled samples for M-Loss from 128 to 512 (ablation)M-TIES · 8-task average, ViT-B/32accuracy (%): 73.232 -> 73.227No clear effect
vary random seed for sampling M-Loss data (ablation)M-TIES · 8-task average, ViT-B/32accuracy (%): 73.225 / 73.165 / 73.232No clear effect
apply M-Loss pruning only to first and last few layers (Few-Layer M-TIES) (ablation)M-TIES · 8-task average, ViT-B/32accuracy (%): 73.232 -> 73.216No clear effect
apply M-Loss pruning only to a few layers (ablation)M-TIES · 8-task average, ViT-L/14accuracy (%): 85.278 -> 85.259No 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