Flask is a great lightweight web framework for Python that makes it really easy to put together an API for a data science model quickly. In this post, I cover some of the basics for getting started with Flask and serving your model with a web API.
Machine learning has a lot of fantastic practical applications, but there comes a point where you want to package up your model so it can be used in practical applications. Web APIs provide a great mechanism to deploy and share models to the wider world.
Sometimes it's useful to split the components of a Python library across multiple respositories, but still access them from the same namespace. Namespace packages are a great mechanism for achieving this.
Managing different versions of the Python interpreter and installed packages can be a pain. pyenv and the extension pyenv-virtualenv are great tools for carefully controlling your development environment.
Neural network training can be a computationally expensive task, especially when dealing with larger datasets. Here I document some optimisation of my neural network implementation to take advantage of the parallel performance on my computer.
Selecting a good optimisation algorithm and good learning parameters is key to efficiently and effectively train our digit characterisation neural network. Here we compare several optimisers empirically.
In previous posts we covered the theory and implementation of neural networks. Now it's time to demonstrate a common application of them - recognition of handwritten digits.
You've written a great Python package and want to share it with the world. But how should you go about that? Here we discuss how to distribute your project in a way that allows other Python users to use your code as easily as possible.