Transformer Layers#

Module

MAdds

FLOPs

Parameters

Memory

torch.nn.Transformer

x

x

✓

x

torch.nn.TransformerEncoder

x

x

✓

x

torch.nn.TransformerDecoder

x

x

✓

x

torch.nn.TransformerEncoderLayer

x

x

✓

x

torch.nn.TransformerDecoderLayer

x

x

✓

x

torch.nn.MultiheadAttention

x

x

✓

x