fail to set batch_first == True

[RiqiangGao]

Thanks for sharing your code. When I use this code, I find that when I set the batch_first == True, the net still with batch_first == False version. Any ideas?

Thanks again

[kamo-naoyuki]

Thank you for using my modules and reporting issue, but I'm not sure your case.
The behaviour of batch_first is very simple and I think there are no bugs.

pytorch_convolutional_rnn/convolutional_rnn/functional.py

Lines 307 to 314 in 094b3bd

	def forward(input, weight, hidden, batch_sizes):
	if batch_first and batch_sizes is None:
	input = input.transpose(0, 1)
	nexth, output = func(input, hidden, weight, batch_sizes)
	if batch_first and batch_sizes is None:
	output = output.transpose(0, 1)

If you are using PackedSequence, this option should be ignored.

[RiqiangGao]

Thank you for using my modules and reporting issue, but I'm not sure your case.
The behaviour of batch_first is very simple and I think there are no bugs.

pytorch_convolutional_rnn/convolutional_rnn/functional.py

Lines 307 to 314 in 094b3bd

def forward(input, weight, hidden, batch_sizes):
if batch_first and batch_sizes is None:
input = input.transpose(0, 1)

 nexth, output = func(input, hidden, weight, batch_sizes) 

 if batch_first and batch_sizes is None: 
     output = output.transpose(0, 1) 

If you are using PackedSequence, this option should be ignored.

Thanks a lot again. I will take a further look.

BTW, I have motivated by your RNN cell functions before and they help my project. And I plan to learn all your functions about RNN. Thanks for sharing.