Facebook's AI Research (FAIR) lab announced Thursday that it's open-sourcing FastText, a library of code for text representation and classification. Additionally, it's publishing its research related to FastText.
In a blog post, FAIR said it was sharing the code and research to "ultimately help us all design better applications and further advances in language understanding". The move could help developers build better bots, which are vulnerable to plenty of flaws but are quickly occupying Facebook's Messenger service. As of late July, Facebook said there were more than 18,000 bots on Messenger.
"With the growing amount of online data, there is a need for more flexible tools to better understand the content of very large datasets, in order to provide more accurate classification results," FAIR's blog post said.
FastText, which is exclusively dedicated to text classification, has proven to be as accurate as deep learning classifiers while "many orders of magnitude" faster for training and evaluation, FAIR said. "We have seen results of models trained on more than 1 billion words in less than 10 minutes using a standard multicore CPU," it wrote. FastText can also classify a half million sentences among more than 300,000 categories in less than five minutes.