AION-1: Omnimodal Foundation Model for Astronomical Sciences
UC Berkeley / Polymathic AI · CNRS · Princeton University Polymathic AI · University of Southern California · University of Cambridge · Axiomatic AI · The Forecasting Company · Ellison Institute for Technology · New York University / Simons Foundation · University of Pennsylvania · Flatiron Institute · Flatiron Institute, Simons Foundation · Presight · IDRIS - CNRS · RIKEN / Flatiron Institute · NVIDIA · New York University · Flatiron Institute, Polymathic AI · University of California, Berkeley · University of Wisconsin--Madison · Genentech / NYU · Polymathic / Simons / NYU
PDF 由论文原始站点提供，PaperCompass 不保存论文文件。
摘要
While foundation models have shown promise across a variety of fields, astronomy lacks a unified framework for joint modeling across its highly diverse data modalities. In this paper, we present AION-1, the first large-scale multimodal foundation family of models for astronomy. AION-1 enables arbitrary transformations between heterogeneous data types using a two-stage architecture: modality-specific tokenization followed by transformer-based masked modeling of cross-modal token sequences. Trained on over 200M astronomical objects, AION-1 demonstrates strong performance across regression, classification, generation, and object retrieval tasks. Beyond astronomy, AION-1 provides a scalable blueprint for multimodal scientific foundation models that can seamlessly integrate heterogeneous combinations of real-world observations. Our model release is entirely open source, including the dataset, training script, and weights.