https://github.com/andreww/keras_play
Playing with Keras
https://github.com/andreww/keras_play
machine-learning python tensorflow
Last synced: about 2 months ago
JSON representation
Playing with Keras
- Host: GitHub
- URL: https://github.com/andreww/keras_play
- Owner: andreww
- Created: 2017-03-28T09:29:27.000Z (over 9 years ago)
- Default Branch: master
- Last Pushed: 2017-03-28T09:38:28.000Z (over 9 years ago)
- Last Synced: 2025-06-17T18:09:48.141Z (about 1 year ago)
- Topics: machine-learning, python, tensorflow
- Language: Jupyter Notebook
- Size: 174 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
Playing around with Keras
Some notebooks as I work out if I can do anything useful
with Keras...
Install
=======
I'm using Keras and the TensorFlow backend. It makes sense to
install this in its own environment and I'm an anaconda user and
am moving most of my python over to python 3. So the first step is
to set up a full new anaconda environment using python 3.5:
conda create -n tensorflow python=3.5 anaconda
Then switch to the new environment:
source activate tensorflow
The next step is to install the TensorFlow backend inside the conda
environment following the documentation at
https://www.tensorflow.org/install/install_mac#installing_with_anaconda
this amounts to running:
pip install --ignore-installed --upgrade https://storage.googleapis.com/tensorflow/mac/cpu/tensorflow-1.0.1-py3-none-any.whl
Do the basc validation by following the instructions at: https://www.tensorflow.org/install/install_mac#ValidateYourInstallation - I get warnings about
tensorflow not using SSE4.1, 4.2 AVX AVX2 and FMA instructions. But I'm only
playing on this machine so I'll take the (probably quite big) hit. Finally,
install keras:
pip install keras
It seems this also installs the theano back end.
Resources
=========
In terms of designing NN this looks useful: http://hagan.okstate.edu/NNDesign.pdf