Caffe2 now supports recurrent neural network (RNN). When Facebook open sourced Caffe2 in April this year, it did not support RNN models. The news was announced by the Caffe2 team via a blog post.
Developed and open sourced by Facebook earlier this year, Caffe2 is an open source, lightweight, modular, and scalable framework to build deep learning based systems. As a matter of fact, Facebook is now using Caffe2 and its deep learning capabilities for its billions of daily translations.
From the blog post:
“Using Caffe2, we significantly improved the efficiency and quality of machine translation systems at Facebook. We got an efficiency boost of 2.5x, which allows us to deploy neural machine translation models into production. As a result, all machine translation models at Facebook have been transitioned from phrase-based systems to neural models for all languages. In addition, several product teams at Facebook, including speech recognition and ads ranking, have started using Caffe2 to train RNN models.”
Developers can implement deep learning based systems using Caffe2 using Python or C++ programming languages for Mac, Ubuntu, CentOS, Windows 10, iOS, and Android operating systems.
Open source project is available on Github.
Here is a list of tutorials: https://caffe2.ai/docs/tutorials