Home BlogDataset Training AI Systems to Code

Training AI Systems to Code

by Morgan Stevens
by
Binary code in green font

IBM has released an open dataset of coding samples, which demonstrate programming tasks, to help train AI systems to write code. The dataset, known as Project CodeNet, includes 14 million code samples in 55 different programming languages. Researchers at IBM have already begun using the dataset to train AI systems to write code and found that the systems achieved a 90 percent accuracy rate in most code classification and code similarity experiments.  

Get the data. 

Image credit: Flickr user Christiaan Colen

You may also like

Show Buttons
Hide Buttons