AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Expert Merging: Model Merging with Unsupervised Expert Alignment and Importance-Guided Layer Chunking

Dengming Zhang, Xiaowen Ma, Zhen-Liang Ni, Zhenkai Wu, Han Shu, Xin Jiang, Xinghao Chen

ICLR 2026 · Large language models › Instruction and fine-tuning · approach: Efficient LLM fine-tuning systems

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 logit alignment (ablation)Expert Merging (InternVL2.5-1B) · Multi-task merging averageaverage score: 54.72 -> 58.11Improved
add logit alignment (ablation)Expert Merging (Mistral-7B) · Multi-task merging averageaverage score: 43.62 -> 47.43Improved
add logit alignment (ablation)Expert Merging (Qwen2-VL-7B) · Multi-task merging averageaverage score: 62.71 -> 63.34Improved
add hidden-state alignment (ablation)Expert Merging (InternVL2.5-1B) · Multi-task merging averageaverage score: 57.75 -> 58.11Improved
add hidden-state alignment (ablation)Expert Merging (Qwen2-VL-7B) · Multi-task merging averageaverage score: 62.43 -> 63.34Improved
add coefficient regularization (ablation)Expert Merging (InternVL2.5-1B) · Multi-task merging averageaverage score: 47.57 -> 58.11Improved
add coefficient regularization (ablation)Expert Merging (Mistral-7B) · Multi-task merging averageaverage score: 40.81 -> 47.43Improved
add coefficient regularization (ablation)Expert Merging (Qwen2-VL-7B) · Multi-task merging averageaverage score: 62.17 -> 63.34Improved
add explicit task weights (ablation)Expert Merging · Multi-task merging averageaverage score: 57.71 -> 58.11 (InternVL)Improved
add importance-guided chunk-wise coefficients (ablation)Expert Merging (Mistral-7B) · Multi-task merging averageaverage score: 47.43 -> 48.71Improved
add importance-guided chunk-wise coefficients (ablation)Expert Merging (InternVL2.5-1B) · Multi-task merging averageaverage score: 58.11 -> 58.45Improved
chunk all layers (2x coefficients) instead of importance-guided (ablation)Expert Merging++ (InternVL2.5-1B) · Multi-task merging averageaverage score: 58.45 -> 56.76Hurt

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