Year: 2025
Type: preprint
Abstract: We investigate when foundation models converge towards shared representations, and how this convergence depends on model capacity, training regime, and model architecture. We take a `science-for-AI' approach, using astronomy as an experimental instrument to test the Platonic Representation Hypothesis and its Aristotelian refinement against an external physical reference. The historical success of astrophysics is evidence that a compact, modality-invariant description of galaxy observables exists, and so representation convergence toward reality should be measurable against the physical parameters astronomers already use. Given this framework, we evaluate eleven foundation model families (spanning classification, self-distillation, joint-embedding prediction, autoencoding, vision-language pre-training, and astro-specific architectures from $\mathcal{O}$(10M)${\to}\mathcal{O}$(10B) parameters) on crossmatched JWST, HSC, and Legacy imagery, and DESI spectroscopy. All models are evaluated frozen, with no astronomy-specific fine-tuning. We probe redshift, stellar mass, and sSFR via linear probes, and local (MKNN) and global (CKA) embedding geometry within families, between modalities, and across architectures. We find that physics performance scales predictably with capacity; probe directions align consistently with expected astrophysical correlations and selection effects; and local (not global) embedding alignment tracks physics performance, including between DESI spectra and HSC imagery---modalities that share essentially no low-level statistics. Our results support the ARH over the strict PRH, demonstrate astronomy's value as an experimental framework for neural representation learning, and suggest that astro-foundation models can build on general-purpose pre-trained architectures, capitalizing on the broader open machine learning community's already-spent computational investment.
Source: arXiv (Cornell University)
Institutions: Sorbonne Université, Institut du Cerveau, Stanford University, University of Hawaiʻi at Mānoa, University of Hawaii System, University of Pittsburgh, Inter-University Centre for Astronomy and Astrophysics, Center for Astrophysics Harvard & Smithsonian, Oak Ridge National Laboratory, Lawrence Berkeley National Laboratory, Space Telescope Science Institute, Johns Hopkins University
Language: English
Subfield: Astronomy and Astrophysics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
Funders: National Science Foundation, U.S. Department of Energy, University of Hertfordshire, University of Illinois at Urbana-Champaign, Office of Science, Science and Technology Facilities Council, Office of Advanced Cyberinfrastructure, Division of Astronomical Sciences, Lawrence Berkeley National Laboratory
Awards: G1044631029, G1079818582, G1143651532, G1286236842, G1453181110, G1813527838, G1863003419, Track 4: Advanced CI Coordination Ecosystem: Monitoring and Measurement Services, G2848635288, G3562386289, Track 1: ACCESS Resource Allocations Marketplace and Platform Services (RAMPS), G4010157413, G419881185, G42916606, G4501827968, ACO: An Open CI Ecosystem to Advance Scientific Discovery (OpenCI), G498139845, Track 3: COre National Ecosystem for CyberinfrasTructure (CONECT), G5287069595, G5367127336, G5464919783, Category I: Bridging the Gap Between AI/ML Computing Demands and Today's Capabilities, G6652958823, G6709186444, G6754955425, G6876926712, Management and Operations of the National Optical Astronomy Observatory, G7354628648, Track 2: Customized Multi-tier Assistance, Training, and Computational Help (MATCH) for End User ACCESS to CI, G8562358136, G8851268641
OpenAlex ID: W7083304308
Cites:
Cited by:
Related to: