/tf-lenet

Migrating from Theano to Tensorflow using Lenet as a case study.

Primary LanguagePythonOtherNOASSERTION

TF-Lenet

MIT License at http://tf-lenet.readthedocs.io/en/latest/license.html Documentation at http://tf-lenet.readthedocs.io/en/latest/ Requirements Status Open Source Love

  • Are you a long time user of Theano and considering migrating to TensorFlow?
  • Are you here to learn what CNNs are and how to implement them in TensorFlow?

Then you're in luck! I myself am just migrating from theano to tensorflow and this is my first implementation in TensorFlow. This is a typical Lenet-5 network trained to classify MNIST-like dataset. This is a simple code that is found in the tutorial that comes with TensorFlow. I made it a little modularized so that it could be re-purposed for other networks later. The documentation website that comes along with this repository helps users migrating from theano to tensorflow.

Obviously, you'd need tensorflow and numpy installed. To run the code simply run:

python main.py

Once run, you can view all summaries at the tensorboard by running:

tensorboard --logdir tensorboard

Refer to the documentation site for more details. Thanks for checking out the code, hope it was useful.

Ragav Venkatesan

http://www.ragav.net