AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

MoMa: A Simple Modular Learning Framework for Material Property Prediction

Botian Wang, Yawen Ouyang, Yaohui Li, Mianzhi Pan, yuanhang tang, Haorui Cui, Yiqun Wang, Jianbing Zhang, Xiaonan Wang, Wei-Ying Ma, Hao Zhou

ICLR 2026 · Probabilistic and scientific modeling › Molecules and proteins · approach: Machine-learned interatomic potentials

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
compose modules with MoMa instead of fine-tuning JMP directlyJMP · Matminer 17 tasksMAE: average improvement of 14.0% over JMP-FT, better on 14 tasksImproved
keep modules separate and compose adaptively instead of multi-task pretraining on all hub tasksJMP · Matminer 17 tasksMAE: 24.8% average margin over JMP-MT, better on 16 of 17 tasksImproved
multi-task pretraining on hub tasks (JMP-MT) versus plain fine-tuning (ablation)JMP · Matminer 17 tasksaverage rank: 3.12 -> 4.53Hurt
use MoMa on the Orb (GNS) backboneOrb-v2 · Matminer 17 tasksMAE: 6.1% average boost, better on 13/17 tasksImproved
AMC weighted selection instead of uniform average of selected modules (ablation)MoMa · Matminer 17 taskstest MAE: 11.0% average MAE increase without weightsImproved
AMC module selection instead of averaging all modules (Model Soup) (ablation)MoMa · Matminer 17 taskstest MAE: 18.0% average MAE increase for All AverageImproved
AMC selection instead of random module selection (ablation)MoMa · Matminer 17 taskstest MAE: 20.2% average MAE increase for Random SelectionImproved
AMC instead of LoRAHub black-box search for composition (ablation)MoMa · Matminer 17 taskstest MAE: 21.8% average MAE reduction, better on 15 tasksImproved
AMC instead of a learned routing network over 18 modules (JMP-(18)) (ablation)MoMa · Matminer 17 taskstest MAE: 15.5% average MAE reduction, better on 17 tasksImproved
AMC convex optimization instead of Softmax Weighting heuristic (ablation)MoMa · Matminer 17 taskstest MAE: 13.7% average MAE reduction, better on 12 tasksImproved
use MoMa in 10-shot settingJMP · Few-shot 10-shotaverage normalized test MAE: 0.7003 -> 0.5503Improved
scale hub from 5 to 30 modules (adding 12 QM9 modules) (ablation)MoMa · Matminer 17 tasksaverage normalized test MAE: 0.2040 -> 0.1759Improved

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