章节2.3.2：位置编码计算结果修正

[EF-Lin233]

2.3.2 位置编码 中的位置示例中的x =[[ 0.1, 0.2, 0.3, 0.4], [0.2, 0.3, 0.4, 0.5], [0.3, 0.4, 0.5, 0.6], [0.4, 0.5, 0.6, 0.7]]
输出结果有误，XPE应该是
[[ 0.100, 1.200, 0.300, 1.400],
[ 1.042, 0.840, 0.410, 1.500],
[ 1.210, -0.016, 0.520, 1.600],
[ 0.541, -0.490, 0.630, 1.700]]] （全部保留三位小数）

测试

class PositionalEncoding(nn.Module):
    def __init__(self, d_model, dropout=0.0, max_len=5000):
        super(PositionalEncoding, self).__init__()
        self.dropout = nn.Dropout(p=dropout)
        pe = torch.zeros(max_len, d_model)
        position = torch.arange(0, max_len, dtype=torch.float).unsqueeze(1)
        div_term = torch.exp(torch.arange(0, d_model, 2).float() * -math.log(10000.0) / d_model)
        pe[:, 0::2] = torch.sin(position * div_term)
        pe[:, 1::2] = torch.cos(position * div_term)
        self.pe = pe.unsqueeze(0)

    def forward(self, x):
        x = x + self.pe[:, : x.size(1)].requires_grad_(False)
        return self.dropout(x)

x = torch.tensor([[ 0.1, 0.2, 0.3, 0.4], [0.2, 0.3, 0.4, 0.5], [0.3, 0.4, 0.5, 0.6], [0.4, 0.5, 0.6, 0.7]])
pe = PositionalEncoding(4)
x = pe(x)
print(x)

得到

tensor([[[ 0.1000,  1.2000,  0.3000,  1.4000],
         [ 1.0415,  0.8403,  0.4100,  1.4999],
         [ 1.2093, -0.0161,  0.5200,  1.5998],
         [ 0.5411, -0.4900,  0.6300,  1.6996]]])