SNACK: A Sequential Notation Framework for Probabilistic Graph Generation
stat.ML, cs.LG
Submitted: 2026-10-14
Updated: 2026-10-18
Comments: 30 pages; Accepted at NeurIPS 2026
Code: https://github.com/leelabsg/snack
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Sources
- Learning Deep Generative Models of Graphs
- MolecularRNN: Generating realistic molecular graphs with optimized properties
- GraphAF: a Flow-based Autoregressive Model for Molecular Graph Generation
- Sequence Level Training with Recurrent Neural Networks
- Synthesizable Molecular Generation via Soft-constrained GFlowNets with Rich Chemical Priors
- Rooted Absorbed Prefix Trajectory Balance with Submodular Replay for GFlowNet Training
- GFlowNets and variational inference
- Stable-GFlowNet: Toward Diverse and Robust LLM Red-Teaming via Contrastive Trajectory Balance
- Beyond Normalization: Rethinking the Partition Function as a Difficulty Scheduler for RLVR
- LLaMA: Open and Efficient Foundation Language Models
- TxGemma: Efficient and Agentic LLMs for Therapeutics
- Robust Scheduling with GFlowNets
- GLU Variants Improve Transformer
- Fast Graph Generation via Spectral Diffusion
- Adam: A Method for Stochastic Optimization
- Molecular Mechanics-Driven Graph Neural Network with Multiplex Graph for Molecular Structures
- Decoupled Weight Decay Regularization
Related papers
- Behavior of prediction performance metrics with rare events
- Optimal Estimation of Generic Dynamics by Path-Dependent Neural Jump ODEs
- A Posterior-Dynamics Framework for Imaging Inverse Problems with Pretrained Diffusion Priors
- Scalable Posterior Uncertainty for Flexible Density-Based Clustering
- A Transport Geometry of Benign Overfitting
- Efficient quadratic entropy with distance sketches