|
from attention.MUSEAttention import MUSEAttention
|
|
import torch
|
|
from torch import nn
|
|
from torch.nn import functional as F
|
|
|
|
|
|
input=torch.randn(50,49,512)
|
|
sa = MUSEAttention(d_model=512, d_k=512, d_v=512, h=8)
|
|
output=sa(input,input,input)
|
|
print(output.shape)
|