Table 1 Architecture of the image encoder and decoder.

From: CalliFormer: a structure-aware transformer for Chinese calligraphy generation

Layer

Encoder

Decoder

Input

1 × 256 × 256

1280 × 1 × 1

L1

64 × 128 × 128

512 × 2 × 2

L2

128 × 64 × 64

512 × 4 × 4

L3

256 × 32 × 32

512 × 8 × 8

L4

512 × 16 × 16

512 × 16 × 16

L5

512 × 8 × 8

256 × 32 × 32

L6

512 × 4 × 4

128 × 64 × 64

L7

512 × 2 × 2

64 × 128 × 128

L8

512 × 1 × 1

1 × 256 × 256