tensoflow函数和pytorch函数之间的转换

tensorflow	pytroch
tf.reshape(input, shape)	input.view()
tf.expand_dims(input, dim)	input.unsqueeze(dim) / input.view()
tf.squeeze(input, dim)	torch.squeeze(dim)/ input.view()
tf.gather(input1, input2)	input1[input2]
tf.tile(input, shape)	input.repeat(shape)
tf.boolean_mask(input, mask)	input[mask] #注意，mask是bool值，不是0,1的数值
tf.concat(input1, input2)	torch.cat(input1, input2)
tf.matmul()	torch.matmul()
tf.minium(input, min)	torch.clamp(input, max=min)
tf.equal(input1, input2)	torch.eq(input1, input2)/ input1 == input2
tf.logical_and(input1, input2)	input1 & input2
tf.logical_not(input) ~	input
tf.reduce_logsumexp(input, [dim])	torch.logsumexp(input, dim=dim)
tf.reduce_any(input, dim)	input.any(dim)
tf.reduce_mean(input)	torch.mean(input)
tf.reduce_sum(input)	input.sum()
tf.transpose(input)	input.t()
tf.softmax_cross_entroy_with_logits(logits, labels)	torch.nn.CrossEntropyLoss(logits, labels)