# How Molecular Generative Models Organize Molecular Identity

arXiv: https://arxiv.org/abs/2608.06956 · HTML: https://mlsanity.com/paper/2608.06956/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 32 references that are in the map.
- First posted on arXiv 7 Aug 2026.

## Builds on (most cited first, 11 of 11)

- [Automatic chemical design using a data-driven continuous representation of molecules](https://mlsanity.com/paper/1610.02415/) (2016)
- [Junction Tree Variational Autoencoder for Molecular Graph Generation](https://mlsanity.com/paper/1802.04364/) (2018)
- [On the Number of Linear Regions of Deep Neural Networks](https://mlsanity.com/paper/1402.1869/) (2014)
- [Score-based Generative Modeling of Graphs via the System of Stochastic Differential Equations](https://mlsanity.com/paper/2202.02514/) (2022)
- [Hierarchical Generation of Molecular Graphs using Structural Motifs](https://mlsanity.com/paper/2002.03230/) (2020)
- [Latent Space Oddity: on the Curvature of Deep Generative Models](https://mlsanity.com/paper/1710.11379/) (2017)
- [On convex decision regions in deep network representations](https://mlsanity.com/paper/2305.17154/) (2023)
- [Characterizing the Latent Space of Molecular Deep Generative Models with Persistent Homology Metrics](https://mlsanity.com/paper/2010.08548/) (2020)
- [SALSA: Semantically-Aware Latent Space Autoencoder](https://mlsanity.com/paper/2310.02744/) (2023)
- [Pointwise Distance Distributions for detecting near-duplicates in large materials databases](https://mlsanity.com/paper/2108.04798/) (2021)
- [MolMiner: Toward Controllable, 3D-Aware, Fragment-Based Molecular Design](https://mlsanity.com/paper/2411.06608/) (2024)

## Abstract (arXiv)

Generative models for matter are often evaluated as samplers over output representations, and their latent spaces are commonly used as proxies for navigating chemical space. Much less is known about how these models internally arrange discrete chemical identities within those representations. We study this arrangement by making molecular identity explicit and pulling it back through the generative process. Through these pullbacks we probe the regions that generate the same object, exposing the trained model's internal repertoire: a fixed partition that determines which objects (novel or not) the model can produce. Across three molecular generative architectures, we find that this repertoire is arranged into piecewise-constant regions separated by recurring coarse-to-fine boundaries. Its organization depends on the representation probed, the identity convention, decoder stochasticity, and the metric used to compare coordinates. During training, local chemical organization stabilizes while the number of distinct molecular identities represented within each neighborhood continues to change. Internal organization must therefore be characterized, rather than assumed, before a generative space can be treated as chemically navigable.