Tensorflow is an opensource library for computation and for large scale machine learning. With Tensorflow you can define a graph in Python and then it runs the code with C++ code. In addition, it will break it up into chunks and run across multiple CPUs or GPUs.
TF.Learn is the Python API (tensorflow.contrib.learn) which is compatible with Scikit-Learn and can be used to setup neural networks.
TF-slim (tensorflow.contrib.slim) is used to build, train, and evaluate neural networks.
Tensorflow automatically takes care of automatic differentiation (or autodiff) by computing the gradients of the function that you define. It also comes with TensorBoard that allows you to visualize the computational graphs, learning curves, etc.
Install tensor flow using pip:
pip install –upgrade tensorflow
For GPU support you need to install tensorflow-gpu instead of tensorflow.
To test the installation output the version of Tensorflow that was installed.
python -c ‘import tensorflow; print(tensorflow.__version__)’