Closed smallbal closed 6 years ago
Thank you for your post. We noticed you have not filled out the following field in the issue template. Could you update them if they are relevant in your case, or leave them as N/A? Thanks. Have I written custom code OS Platform and Distribution TensorFlow installed from TensorFlow version Bazel version CUDA/cuDNN version GPU model and memory Exact command to reproduce
Thanks for replying to my issue, I have updated my comment and added more details.
@petewarden can you please take a look?
Nagging Assignee @bignamehyp: It has been 14 days with no activity and this issue has an assignee. Please update the label and/or status accordingly.
Nagging Assignee @petewarden: It has been 14 days with no activity and this issue has an assignee. Please update the label and/or status accordingly.
Thanks for spotting this! It does look like a bug, sorry about that. Since we've been using that conv model as a baseline (for example in https://arxiv.org/abs/1804.03209 ) it might be awkward to fix it at this stage. I would welcome a PR with a comment added noting this issue, and possibly an alternative conv_with_maxpool implementation if it does produce noticeably better results. Closing for now, but please refer to this if you do get a chance to work on that.
There may be a mistake I found in
tensorflow/tensorflow/example/speech_commands/models.py
: The functioncreate_conv_model()
is to create a "cnn-trad-fpool3" network with 2 max pool layers, but the second max pool layer is missing now. The code can run normally, but I'm not sure if it is the right "cnn-trad-fpool3" network.I put part of the code of
tensorflow/tensorflow/exmaple/speech_commands/models.py
here below and added some comments to explain my points.System information
N/A
v1.9.0-rc0