mlsanityData to 14 Oct 2026

Mol-JEPA: A multimodal Joint Embedding Predictive Architecture for Molecules

arXiv 2608.22642 ↗First posted 23 Aug 2026cs.LG cs.AIcited by 2

Not on the map. Why?

Abstract (arXiv)

Despite recent advances in molecular foundation models, several limitations remain, such as chemically invalid augmentations, modality collapse, and incomplete representation of biochemical environments. To address these challenges, we present Mol-JEPA, a scalable framework for learning molecular world models. Rather than relying on suboptimal molecular perturbations, our model uses modality masking to exploit information from molecular structures, cellular phenotypes, binding affinities, ADMET profiles, quantum chemistry simulations and other drug discovery data. Across various benchmarks, we show that the representations learned by Mol-JEPA deliver strong performance, demonstrating the value of incorporating biochemical context through latent space prediction.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 30 of its 80 references that are in the map. First posted on arXiv 23 Aug 2026.

Builds on

30 of 80 references.

References · 30

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  3. 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,361
  4. 2019Optuna: A Next-generation Hyperparameter Optimization Frameworkcited by 2,501
  5. 2016Automatic chemical design using a data-driven continuous representation of moleculescited by 765
  6. 2022Mind the Gap: Understanding the Modality Gap in Multi-modal Contrastive Representation Learningcited by 561
  7. 2020Masked Label Prediction: Unified Message Passing Model for Semi-Supervised Classificationcited by 506
  8. 2021Understanding Dimensional Collapse in Contrastive Self-supervised Learningcited by 378
  9. 2020ChemBERTa: Large-Scale Self-Supervised Pretraining for Molecular Property Predictioncited by 337
  10. 2025TabICL: A Tabular Foundation Model for In-Context Learning on Large Datacited by 289
  11. 2021Therapeutics Data Commons: Machine Learning Datasets and Tasks for Drug Discovery and Developmentcited by 264
  12. 2025LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristicscited by 254
18 more
  1. 2021ChemRL-GEM: Geometry Enhanced Molecular Representation Learning for Property Predictioncited by 204
  2. 2022Graph neural networks for materials science and chemistrycited by 182
  3. 2025UMA: A Family of Universal Models for Atomscited by 105
  4. 2022Does GNN Pretraining Help Molecular Representation?cited by 60
  5. 2023MolFM: A Multimodal Molecular Foundation Modelcited by 50
  6. 2023Enhancing Activity Prediction Models in Drug Discovery with the Ability to Understand Human Languagecited by 44
  7. 2023Towards Foundational Models for Molecular Learning on Large-Scale Multi-Task Datasetscited by 39
  8. 2025Joint Embedding vs Reconstruction: Provable Benefits of Latent Space Prediction for Self Supervised Learningcited by 36
  9. 2020GraphCL: Contrastive Self-Supervised Learning of Graph Representationscited by 26
  10. 2025Benchmarking Pretrained Molecular Embedding Models For Molecular Representation Learningcited by 15
  11. 2025Deep Learning Foundation Models from Classical Molecular Descriptorscited by 14
  12. 2025stable-pretraining-v1: Foundation Model Research Made Simplecited by 10
  13. 2023Exploring QSAR Models for Activity-Cliff Predictioncited by 9
  14. 2024WelQrate: Defining the Gold Standard in Small Molecule Drug Discovery Benchmarkingcited by 5
  15. 2024MolMix: A Simple Yet Effective Baseline for Multimodal Molecular Representation Learningcited by 3
  16. 2026Do Larger Models Really Win in Drug Discovery? A Benchmark Assessment of Model Scaling in AI-Driven Molecular Property and Activity Predictioncited by 2
  17. 2025Joint Embedding Predictive Architecture for self-supervised pretraining on polymer molecular graphscited by 1
  18. 2026When to Align, When to Predict: A Phase Diagram for Multimodal Learningcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Can phenotypic activity be predicted without experimental readouts?cited by 0
  2. 2026Atom-JEPA: Joint-Embedding Predictive Architecture for 3D Atomistic Systemscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026MolEmb: Multimodal Large Language Models Can Be Strong Molecular Embedding Models
  2. 2025Unified Molecule Pre-training with Flexible 2D and 3D Modalities: Single and Paired Modality Integration
  3. 2025Mol-LLM: Multimodal Generalist Molecular LLM with Improved Graph Utilization
  4. 2026VQ-Atom: Semantic Discretization of Local Atomic Environments for Molecular Representation Learning
  5. 2023Beyond Chemical Language: A Multimodal Approach to Enhance Molecular Property Prediction
  6. 2026What Does a Chemical Language Model Know About Molecules?
  7. 2024MOL-Mamba: Enhancing Molecular Representation with Structural & Electronic Insights
  8. 2025Thinking like a CHEMIST: Combined Heterogeneous Embedding Model Integrating Structure and Tokens