我试图将一些pytorch代码导入到tensorflow中,我知道torch.nn.functional.conv1d()是tf.nn.conv1d(),但是我担心tf的版本中仍然存在一些差异。具体来说,我在tf.conv1d中找不到组参数。例如:以下代码输出两个不同的结果:
火炬:
inputs = torch.Tensor([[[1, 1, 1, 1],[2, 2, 2, 2],[3, 3, 3, 3]]]) #batch_sizex seq_length x embed_dim,
inputs = inputs.transpose(2,1) #batch_size x embed_dim x seq_length
batch_size, embed_dim, seq_length = inputs.size()
kernel_size = 3
in_channels = 2
out_channels = in_channels
weight = torch.ones(out_channels, 1, kernel_size)
inputs = inputs.contiguous().view(-1, in_channels, seq_length) #batch_size*embed_dim/in_channels x in_channels x seq_length
inputs = F.pad(inputs, (kernel_size-1,0), 'constant', 0)
output = F.conv1d(inputs, weight, padding=0, groups=in_channels)
output = output.contiguous().view(batch_size, embed_dim, seq_length).transpose(2,1)
输出:
tensor([[[1., 1., 1., 1.],
[3., 3., 3., 3.],
[6., 6., 6., 6.]]])
Tensorflow:
inputs = tf.constant([[[1, 1, 1, 1],[2, 2, 2, 2],[3, 3, 3, 3]]], dtype=tf.float32) #batch_sizex seq_length x embed_dim
inputs = tf.transpose(inputs, perm=[0,2,1])
batch_size, embed_dim, seq_length = inputs.get_shape()
print(batch_size, seq_length, embed_dim)
kernel_size = 3
in_channels = 2
out_channels = in_channels
weight = tf.ones([kernel_size, in_channels, out_channels])
inputs = tf.reshape(inputs, [(batch_size*embed_dim)//in_channels, in_channels, seq_length], name='inputs')
inputs = tf.transpose(inputs, perm=[0, 2, 1])
padding = [[0, 0], [(kernel_size - 1), 0], [0, 0]]
padded = tf.pad(inputs, padding)
res = tf.nn.conv1d(padded, weight, 1, 'VALID')
res = tf.transpose(res, perm=[0, 2, 1])
res = tf.reshape(res, [batch_size, embed_dim, seq_length])
res = tf.transpose(res, perm=[0, 2, 1])
print(res)
输出:
[[[ 2. 2. 2. 2.]
[ 6. 6. 6. 6.]
[12. 12. 12. 12.]]], shape=(1, 3, 4), dtype=float32)
答案 0 :(得分:2)
这些版本之间没有差异,您只是在设置不同的内容。为了获得与Tensorflow中完全相同的结果,请将指定权重的行更改为:
weight = torch.ones(out_channels, 2, kernel_size)
,因为您的输入有两个输入通道,正如您在TF中正确声明的那样:
weight = tf.ones([kernel_size, in_channels, out_channels])
您误解了groups
中负责什么的pytorch
参数。它限制了每个过滤器使用的通道数(在这种情况下,只有2个input_channels
除以2才能得到一个)。
有关2D
卷积的更直观说明,请参见here。