Machine Learning Assists Python Code Completion
Kite — the new programming assistant that suggests code for Python based on data collected online.
Kite is the new program developed to assist Python users by using machine learning to complete code in rapid time. Kite is moving beyond the old-school auto-suggest algorithms used by other code completing systems and is now available for use in most of the major code editors including PyCharm, IntelliJ, Sublime Text, Atom, Vim and Microsoft Visual Studio Code.
At the time of writing Kite is only supporting Python by there are moves afoot by developers to start to support other languages in the very near future.
Kite’s code completion is different because it scans Python code which is publicly available on GitHub. The Kite model uses machine learning and is focused on the syntax trees that come from the code rather than the actual code text. By looking at these trees Kite can ascertain the intent and context behind the code making it able to deliver automatic suggestion and completion of code based on what code has been designed and written for the in past.
The latest Kite release has incorporated expanded code suggestion and clearly demonstrates the value of this approach. Although previous versions gave some useful suggestions like the next likely token this latest version can suggest an entire function call including the available arguments and meanings.
The new version also means that Kite can be deployed both on a server and a local system which is in line with current developments in machine learning where models are becoming smaller and easier to deploy even where the CPU capabilities and storage are modest.
If you’re looking for a company that provides ERP Integration and other software development services contact us.