The Complete Magazine on Open Source

Facebook launches open source Torchnet toolkit to enable new AI experiments

, / 1920 0


Intending to take artificial intelligence (AI) to new levels, Facebook has launched Torchnet. The open source software toolkit is based on deep machine learning framework Torch and is aimed to foster “rapid and collaborative development” of deep learning developments by the Torch community.

Debuted at the International Conference on Machine Learning (ICML) in New York, Torchnet uses AI concepts through a series of boilerplate code, key abstractions and reference implementations. The offerings through the toolkit can be combined or reuse separately to uplift the developments among deep learning researchers.

Facebook Torchnet

Facebook’s Torchnet offers new experiences to deep learning researchers

“We created Torchnet to give researchers clear guidelines on how to set up their code, and boilerplate code that helps them develop more quickly. The modular Torchnet design makes it easy to test a series of coding variants focused around the data set, the data loading process, and the model, as well as optimisation and performance measures,” Torchnet team led by Facebook’s Artificial Research (FAIR) lab writes in a blog post.

Torchnet is written in multi-paradigm programming language, Lua. It is designed to run on any devices with a standard x86 chip.

Unlike traditional deep learning frameworks

Facebook’s research scientists consider that Torchnet is “substantially different” from deep learning frameworks such as Chainer, TensorFlow and Theano as it provides a framework layer on top of a native deep learning framework. This comes contrary to the traditional model that primarily focuses on performing efficient inference and gradient computing in some deep networks.

Facebook is not the only tech company that has started viewing deep learning as its future aspects. Starting from Amazon to Google and Microsoft, all the leading players are in the race to expand their presence in the world of AI with their offerings. In fact, Twitter is already using the same Torch framework to build machine learning software.

Deep learning on new levels

The growth of deep learning tools like Torchnet will certainly reach new heights in the coming future. However, it is difficult to say how Facebook and other tech giants will utilise this emerging technology trend to expand their coverage and flatter some initial AI models.