Zatom-2: Multitask Pretraining on Atomistic Data for Generative Modeling across Domains
cs.LG, cs.AI
Submitted: 2026-10-08
Updated: 2026-10-10
Comments: Corrected project page URL to https://zatom-ai.github.io/nucleus
Code: https://github.com/Zatom-AI/nucleus
Project page: https://zatom-ai.github.io/nucleus
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- LeMat-GenBench: A Unified Evaluation Framework for Crystal Generative Models
- Learning Inter-Atomic Potentials without Explicit Equivariance
- Crystalite: A Lightweight Transformer for Efficient Crystal Modeling
- Elucidating the Design Space of Diffusion-Based Generative Models
- The Open Molecules 2025 (OMol25) Dataset, Evaluations, and Models
- Generating Novel, Designable, and Diverse Protein Structures by Equivariantly Diffusing Oriented Residue Clouds
- Geometric Flow Matching for Molecular Conformation Generation via Manifold Decomposition
- Zatom-1: Towards a Multimodal Foundation Model for 3D Molecules and Materials
- Orb: A Fast, Scalable Neural Network Potential
- Self-Conditioned Denoising for Atomistic Representation Learning
- SimpleFold: Folding Proteins is Simpler than You Think
- Emyx: Fast and efficient all-atom protein generation
- CrystalREPA: Transferring Physical Priors from Universal MLIPs to Crystal Generative Models
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks