Numba series part 2: Custom data types and parallelization
Here we will focus on how we can use custom data types inside of Numba optimized functions as well as parallelization.
Numba series part 1: The @jit decorator and some more Numba basics
In this part we'll have a closer look at the @jit decorator of the Numba library and talk about some pitfalls, as well as some more basics.
Introduction to the Numba library
The Numba library allows you to achieve near C/C++/Fortran performance with your Python code without many code changes. This post will introduce the concept of Numba and compare the actual performance gain.
Speeding up TensorFlows Input Pipeline
Doubling the training speed by adding two arguments to the new input pipeline - or why you should always carefully read the docs.
Example of TensorFlows new Input Pipeline
With version 1.2rc0 TensorFlow has gotten a new input pipeline. In this blog post I will explain usage and give an example of an entire input pipeline.
Finetuning AlexNet with TensorFlow
This blog post will guide you on how to finetune AlexNet with pure TensorFlow.
Understanding the backward pass through Batch Normalization Layer
An explanation of gradient flow through BatchNorm-Layer following the circuit representation learned in Standfords class CS231n.