Mol-JEPA: multimodal joint-embedding predictive architecture hits SOTA on molecule benchmarks
randall_balestr · x · 2026-08-26
A new arXiv paper, "Mol-JEPA: A multimodal Joint Embedding Predictive Architecture for Molecules" (Rottach, Balestriero, et al.), proposes a scalable framework for learning molecular world models. It targets three persistent limitations of molecular foundation models: chemically invalid augmentations, modality collapse, and incomplete representation of biochemical environments.
Instead of suboptimal molecular perturbations, Mol-JEPA uses modality masking to jointly exploit molecular structures, cellular phenotypes, binding affinities, ADMET profiles, quantum chemistry simulations, and other drug discovery data via latent-space prediction. The learned representations deliver SOTA performance across multiple benchmarks, and the paper provides numerous practical guidelines for multi-modality (>2) pretraining from scratch.
Related event: Mol-JEPA: Molecular Foundation Model Learning Across 14 Modalities(2 posts)→
More from Research
- Christian Szegedy predicts AI will tear through math, then physics, then biology — ChrSzegedy · 2026-09-23
- Researcher Slams Journal Peer Review for Ignoring SOTA Bias-Correction Method ChromBPNet — anshulkundaje · 2026-09-23
- Stanford's Kundaje Calls Out TRACEBIND Paper for Ignoring ChromBPNet Baseline — anshulkundaje · 2026-09-23
- Academia operates like a content farm amid frontier-lab science disruption, researcher argues — RexDouglass · 2026-09-23
- EvolveTrade Paper: Evolving Agent System Prompts Beats Fixed-Prompt Trading Baselines — dair_ai · 2026-09-23
- MBZUAI x Ant Group open-source OPD work: IER selector helps thinking mode too — jiank_uiuc · 2026-09-23