ChargeFlow dataset release: training source corpora, split manifests, and external periodic benchmark

Read full text Summarise with VIRAT
Authors: Thin Nguyen, Sherif Abdulkader Tawfik, Truyen Tran, Svetha Venkatesh
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Dataset release accompanying the manuscript ChargeFlow: Flow-Matching Refinement of Charge-Conditioned Electron Densities. The package contains two Materials Project source-corpus archives used to define the manuscript training and internal-development-holdout splits, sanitized split manifests that map exact paper samples onto archive members, and an external periodic benchmark spanning perovskites, charged defects, multisite and special diamond defects, MOFs, extreme MOFs, organic crystals, and extreme organic crystals. The package-level README, archive inventory, and split manifests describe how the archives are organized. The internal development hold-out split is included for reproducibility of model development and is not the manuscript's external benchmark for out-of-distribution evaluation.

Fields of Study & Subjects

Machine Learning in Materials Science Advanced Chemical Physics Studies Computational Drug Discovery Methods

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
25 March 2026
Language
EN
Document type
Dataset