A controlled head-to-head study shows Vision Transformers beat convolutional networks on real X-ray diffraction data once fine-tuned, while CNNs remain more robust for zero-shot transfer.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results