aiqbench analysis
This study demonstrates that foundation models pretrained on high-energy collider data effectively transfer to low-energy neutrino experiments, significantly outperforming models trained from scratch or initialized with unrelated text. The findings validate the utility of detector-agnostic inference in particle physics, suggesting that specialized AI architectures can generalize across vast differences in energy scales and experimental setups. This marks a pivotal shift toward leveraging foundation models to accelerate discovery in high-energy physics without requiring massive new compute budgets for every new experiment.
Cross-Domain Transfer with Particle Physics Foundation Models: From Jets to Neutrino Interactions
arXiv:2604.12364v2 Announce Type: replace-cross Abstract: Future AI-based studies in particle physics will likely start from a foundation model to accelerate training and enhance sensitivity. As a step toward a general-purpose foundation model for particle physics, we investigate whether the OmniLearned and ParticleViT foundation models pretrained on diverse high-$Q^2$ simulated and real $pp$ and $ep$ collisions retain useful knowledge to a few-GeV fixed-target neutrino experiment. We process MINERvA neutrino--nucleus scattering events and evaluate pretrained models on two types of tasks: regression of available energy and binary classification of charged-current pion final states ($\mathrm{CC1\pi^{\pm}}$, $\mathrm{CCN\pi^{\pm}}$, and $\mathrm{CC1\pi^{0}}$). Pretrained OmniLearned and ParticleViT models outperform similarly sized models trained from scratch at the same compute budget, with the largest gains for OmniLearned on regression and for ParticleViT on classification. When the same transformer architecture is instead initialized from unrelated text pretraining (BERT), this advantage appears only marginally for classification in terms of compute efficiency and not in any way for regression. These results suggest that particle-level foundation models acquire inductive biases that generalize across large differences in energy scale, detector technology, and underlying physics processes, pointing toward detector-agnostic inference in particle physics.
Abstract:Future AI-based studies in particle physics will likely start from a foundation model to accelerate training and enhance sensitivity. As a step toward a general-purpose foundation model for particle physics, we investigate whether the OmniLearned and ParticleViT foundation models pretrained on diverse high-$Q^2$ simulated and real $pp$ and $ep$ collisions retain useful knowledge to a few-GeV fixed-target neutrino experiment. We process MINERvA neutrino--nucleus scattering events and evaluate pretrained models on two types of tasks: regression of available energy and binary classification of charged-current pion final states ($\mathrm{CC1\pi^{\pm}}$, $\mathrm{CCN\pi^{\pm}}$, and $\mathrm{CC1\pi^{0}}$). Pretrained OmniLearned and ParticleViT models outperform similarly sized models trained from scratch at the same compute budget, with the largest gains for OmniLearned on regression and for ParticleViT on classification. When the same transformer architecture is instead initialized from unrelated text pretraining (BERT), this advantage appears only marginally for classification in terms of compute efficiency and not in any way for regression. These results suggest that particle-level foundation models acquire inductive biases that generalize across large differences in energy scale, detector technology, and underlying physics processes, pointing toward detector-agnostic inference in particle physics.
High Energy Physics - Experiment (hep-ex); Machine Learning (cs.LG); High Energy Physics - Phenomenology (hep-ph); Data Analysis, Statistics and Probability (physics.data-an)
arXiv:2604.12364 [hep-ex]
(or arXiv:2604.12364v2 [hep-ex] for this version)
Full text under the publisher's open licence · 212 words
Related coverage
Pre-Training for Simulation-Based Science: A Study on Jet Foundation Model Training Objectives
arXiv · Jun 16, 2026
News
Quantum Computing Report · Jul 25, 2026
Pre-Training for Simulation-Based Science: A Study on Jet Foundation Model Training Objectives
arXiv · Aug 20, 2026
[AINews] Reality Checks on AI News (Yegge shuts down Gas Town, Databricks’ +60% Astra cost)
Latent Space · Sep 17, 2026