#相变

This research reveals how model architecture and data statistics jointly determine representation efficiency in large models. The findings help explain why wider models can represent more features without proportional loss increases, impacting how we scale AI systems. For developers and researchers working with large language models, these insights can guide architecture decisions and data preprocessing strategies to optimize representation capacity.

This research introduces TetrisCNN, a convolutional architecture with differently shaped filters, to detect phases of matter in quantum simulator data. Applied to experimental snapshots of two-dimensional Ising and XY quantum simulators, the model identifies phase transitions and crossovers while expressing its latent representation and decision boundaries as symbolic formulas built from experimentally measurable spin correlators.

For researchers in physics or machine learning, this provides a new method for sampling in high-dimensional systems, particularly at phase transitions.