Our attempt at extracting everything ever said about every galaxy. GalaxyWiki brings together more than 44,000 discussions of 13,616 galaxies, with excerpts and links to the original papers.
Identifying Scientifically Interesting Galaxies with Vision-Language Models
Yixuan Cheng, Nolan Koblischke, Jo Bovy
Machine Learning and the Physical Sciences (ML4PS 2026)
We ranked Hubble images by scientific interest with vision-language models. Gemini recovered 98.2% of 167 reference anomalies in the 2,095 highest-scoring images.
AION-Search: Semantic Search for 100M+ Galaxy Images Using AI-Generated Captions
Nolan Koblischke, Liam Parker, Francois Lanusse, Irina Espejo Morales, Jo Bovy, Shirley Ho
A text-based search engine for galaxy images trained on AI-generated captions. Enables flexible semantic search for rare phenomena without manual labeling.
Gravity-Bench-v1: A Benchmark on Gravitational Physics Discovery for Agents
Nolan Koblischke, Hyunseok Jang, Kristen Menou, Mohamad Ali-Dib
A benchmark that challenges AI agents on gravitational physics discovery tasks, evaluating scientific reasoning and observational planning capabilities within controlled, physics-based environments.
AstroCoder: AI-Powered Discovery and Usage of Astronomy Software
Transforming how astronomers discover code: AI-powered search across 2,270 astronomy tools. From finding the right package to generating working examples in seconds.
SpectraFM: Tuning into Stellar Foundation Models
Nolan Koblischke, Jo Bovy
NeurIPS 2024 Workshop: Foundation Models for Science
Discovered that pulsation impacts the TRGB in the SMC as in the LMC, with older, longer period pulsators leading to a highly accurate TRGB tip measurement.
Small-amplitude Red Giants Elucidate the Nature of the Tip of the Red Giant Branch as a Standard Candle
Richard I. Anderson, Nolan Koblischke, Laurent Eyer
Discovered that red giant pulsation adds a new dimension of the TRGB method, leading to an enhanced method for measuring the Universe's expansion rate.