search

chainer / chainermn

ChainerMN: Scalable distributed deep learning with Chainer
https://chainer.org
MIT License
207 stars 57 forks source link

Error with imagenet example for CPU #125

Closed pbalapra closed 7 years ago

pbalapra commented 7 years ago

I am trying to modify imagenet example to run on CPUs. After couple of modifications, the code starts but immediately exits with the following error.

File "train_imagenet.py", line 80, in get_example image -= self.mean[:, top:bottom, left:right] ValueError: non-broadcastable output operand with shape (1,227,227) doesn't match the broadcast shape (3,227,227)

Could you please let me know how to address this!?

keisukefukuda commented 7 years ago

Hi @pbalapra , Do you mind providing the code you are working on? We haven't experienced such error you described. From the message you provided, the error seems to be related to the modification you made.

Does the error occur in non-ChainerMN version of the ImageNet example from Chainer repository? It would be helpful to start from it to simplify the problem.

pbalapra commented 7 years ago

Thanks for your reply.

Please see the following URL for the modified code: https://gist.github.com/pbalapra/4b0a79d283a6b311fe2103718f261490

pbalapra commented 7 years ago

@keisukefukuda Could you please let me know if you had a chance to look at this!? Thanks..

iwiwi commented 7 years ago

Does it also happen with the original Chainer example? Our ImageNet example is heavily based on Chainer example https://github.com/chainer/chainer/blob/master/examples/imagenet/train_imagenet.py , and so, if it happens without ChainerMN part, then this issue should be moved to Chainer repository. Thanks.

iwiwi commented 7 years ago

Because of no response, we once close this issue. Please reopen when you come back. Thanks!