mlsanityData to 7 Oct 2026

LeMat-GenBench: A Unified Evaluation Framework for Crystal Generative Models

arXiv 2512.04562 ↗First posted 4 Dec 2025cs.LGcited by 20

Not on the map. Why?

Abstract (arXiv)

Generative machine learning (ML) models hold great promise for accelerating materials discovery through the inverse design of inorganic crystals, enabling an unprecedented exploration of chemical space. Yet, the lack of standardized evaluation frameworks makes it challenging to evaluate, compare, and further develop these ML models meaningfully. In this work, we introduce LeMat-GenBench, a unified benchmark for generative models of crystalline materials, supported by a set of evaluation metrics designed to better inform model development and downstream applications. We release both an open-source evaluation suite and a public leaderboard on Hugging Face, and benchmark 12 recent generative models. Results reveal that an increase in stability leads to a decrease in novelty and diversity on average, with no model excelling across all dimensions. Altogether, LeMat-GenBench establishes a reproducible and extensible foundation for fair model comparison and aims to guide the development of more reliable, discovery-oriented generative models for crystalline materials.

Key facts. Cited by 20 papers in the map (data through 7 Oct 2026). Builds on 49 of its 95 references that are in the map. First posted on arXiv 4 Dec 2025.

Builds on

49 of 95 references.

References · 49

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  3. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  4. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  5. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  6. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  7. 2022Flow Matching for Generative Modelingcited by 5,426
  8. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,550
  9. 2022Building Normalizing Flows with Stochastic Interpolantscited by 964
  10. 2017A Brief Survey of Deep Reinforcement Learningcited by 466
  11. 2021Flow Network based Generative Models for Non-Iterative Diverse Candidate Generationcited by 384
  12. 2022The Vendi Score: A Diversity Evaluation Metric for Machine Learningcited by 299
37 more
  1. 2021GFlowNet Foundationscited by 286
  2. 2021Crystal Diffusion Variational Autoencoder for Periodic Material Generationcited by 268
  3. 2023Language Modeling Is Compressioncited by 225
  4. 2020Machine Learning Force Fieldscited by 204
  5. 2018Fr'echet ChemNet Distance: A metric for generative models for molecules in drug discoverycited by 191
  6. 2022Sample Efficiency Matters: A Benchmark for Practical Molecular Optimizationcited by 151
  7. 2023Crystal Structure Prediction by Joint Equivariant Diffusioncited by 142
  8. 2024FlowMM: Generating Materials with Riemannian Flow Matchingcited by 105
  9. 2025UMA: A Family of Universal Models for Atomscited by 99
  10. 2024A Comprehensive Survey of Scientific Large Language Models and Their Applications in Scientific Discoverycited by 96
  11. 2022Forces are not Enough: Benchmark and Critical Evaluation for Machine Learning Force Fields with Molecular Simulationscited by 92
  12. 2024Fine-Tuned Language Models Generate Stable Inorganic Materials as Textcited by 92
  13. 2025All-atom Diffusion Transformers: Unified generative modelling of molecules and materialscited by 75
  14. 2024Space Group Constrained Crystal Generationcited by 71
  15. 2023A Hitchhiker's Guide to Geometric GNNs for 3D Atomic Systemscited by 65
  16. 2023Generative Flow Networks as Entropy-Regularized RLcited by 60
  17. 2023Language models can generate molecules, materials, and protein binding sites directly in three dimensions as XYZ, CIF, and PDB filescited by 59
  18. 2020An invertible crystallographic representation for general inverse design of inorganic crystals with targeted propertiescited by 53
  19. 2023Scalable Diffusion for Materials Generationcited by 52
  20. 2024FlowLLM: Flow Matching for Material Generation with Large Language Models as Base Distributionscited by 52
  21. 2025SymmCD: Symmetry-Preserving Crystal Generation with Diffusion Modelscited by 48
  22. 2024Discrete Probabilistic Inference as Control in Multi-path Environmentscited by 40
  23. 2024Space Group Informed Transformer for Crystalline Materials Generationcited by 40
  24. 2022Physics Guided Deep Learning for Generative Design of Crystal Materials with Symmetry Constraintscited by 30
  25. 2023Crystal-GFN: sampling crystals with desirable properties and constraintscited by 29
  26. 2023Towards Symmetry-Aware Generation of Periodic Materialscited by 25
  27. 2025Wyckoff Transformer: Generation of Symmetric Crystalscited by 24
  28. 2023Multimodal Foundation Models for Material Property Prediction and Discoverycited by 19
  29. 2024Foundational Large Language Models for Materials Researchcited by 18
  30. 2025Periodic Materials Generation using Text-Guided Joint Diffusion Modelcited by 15
  31. 2025Space Group Equivariant Crystal Diffusioncited by 15
  32. 2025PLaID++: A Preference Aligned Language Model for Targeted Inorganic Materials Designcited by 11
  33. 2025Space Group Conditional Flow Matchingcited by 7
  34. 2023Discovery of Novel Reticular Materials for Carbon Dioxide Capture using GFlowNetscited by 6
  35. 2025deCIFer: Crystal Structure Prediction from Powder Diffraction Data using Autoregressive Language Modelscited by 5
  36. 2025LLM-Fusion: A Novel Multimodal Fusion Model for Accelerated Material Discoverycited by 3
  37. 2025Catalyst GFlowNet for electrocatalyst design: A hydrogen evolution reaction case studycited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 1; 2026: 19.

20 papers.

Sort:

By area:

  1. 2026Crystalite: A Lightweight Transformer for Efficient Crystal Modelingcited by 9
  2. 2026Open Materials Generation with Inference-Time Reinforcement Learningcited by 6
  3. 2026MADE: Benchmark Environments for Closed-Loop Materials Discoverycited by 5
  4. 2026Multimodal Crystal Flow: Any-to-Any Modality Generation for Unified Crystal Modelingcited by 4
  5. 2025Transport Novelty Distance: A Distributional Metric for Evaluating Material Generative Modelscited by 2
  6. 2026Substitution-Based Analysis of Structural Novelty for Generative Models of Materialscited by 2
  7. 2026MatMind: A Structure-Activity Knowledge-Driven Generative Foundation Model for Materials Sciencecited by 1
  8. 2026OMatG-flash: An All-Atom Flow Map with Reinforce Adjoint Matching for Scalable Materials Discoverycited by 1
  9. 2026Inverse Design of Inorganic Compounds with Generative AIcited by 0
  10. 2026PRISMat: Policy-Driven, Permutation-Invariant Autoregressive Material Generationcited by 0
  11. 2026Flowing with Confidencecited by 0
  12. 2026Surrogate-Gated Generation and Foundation-Model Embeddings for Bayesian Materials Designcited by 0
8 more
  1. 2026Chemical filters for ultra-high-throughput materials screening and generationcited by 0
  2. 2026Representations from Pretrained Machine-Learning Interatomic Potentials as Coarse Coordinates for Material Generation and Evaluationcited by 0
  3. 2026DynaCrys: Crystal Generation with Dynamic Space-Group Diffusioncited by 0
  4. 2026Autonomous discovery of new structure-plausibility laws for explainable and rapid crystal diagnosis and screeningcited by 0
  5. 2026Deep Generative Crystal Structure Prediction: A Benchmark Study and a Controlled Test of Prototype Dependencecited by 0
  6. 2026Let CSP Be Your ANCHOR: Adaptive Crystal Search over Frozen Structure Priorscited by 0
  7. 2026GLASS: Global Latent Aggregation with Slot-based Set Decoding for Scalable All-Atom Crystal Generationcited by 0
  8. 2026Zatom-2: Multitask Pretraining on Atomistic Data for Generative Modeling across Domainscited by 0