python - 创建变压器无峰值掩码时如何修复 numpy 中的 "TypeError: data type not understood"

Question

我正在尝试通过博客文章实现和训练 NMT 变压器，一切正常，除了我无法创建无峰值掩模，因为我收到此错误:“TypeError:数据类型不理解”

代码:

target_seq = batch.Python.transpose(0,1)
target_pad = PY_TEXT.vocab.stoi['<pad>']
target_msk = (target_seq != target_pad).unsqueeze(1)
size = target_seq.size(1) # get seq_len for matrix
nopeak_mask = np.triu(np.ones(1, size, size),
k=1).astype('uint8')
nopeak_mask = Variable(torch.from_numpy(nopeak_mask) == 0)
target_msk = target_msk & nopeak_mask

错误消息:

TypeError                                 Traceback (most recent call last)
<ipython-input-36-e19167b74ba0> in <module>()
      4 target_msk = (target_seq != target_pad).unsqueeze(1)
      5 size = target_seq.size(1) # get seq_len for matrix
----> 6 nopeak_mask = np.triu(np.ones(1, size, size),
      7 k=1).astype('uint8')
      8 nopeak_mask = Variable(torch.from_numpy(nopeak_mask) == 0)

~/.local/lib/python3.6/site-packages/numpy/core/numeric.py in ones(shape, dtype, order)
    201 
    202     """
--> 203     a = empty(shape, dtype, order)
    204     multiarray.copyto(a, 1, casting='unsafe')
    205     return a

TypeError: data type not understood

Answer 1

np.triu 的第一个输入应该是所需大小的元组，而不是 numpy 数组。

尝试:

np.triu((1, size, size), k=1).astype("uint8")