PF-ViT: Parallel and Fast Vision Transformer for Offline Handwritten Chinese Character Recognition

Figure 6

The visualization of the validation accuracy of the T-ViT model (3 encoders per channel).