Facebook makes damn fine Caffe2 open source to encourage in-app machine learning

News

Facebook has made its lightweight machine learning Caffe2 project open-source as it looks to bring the technology to mobile devices.

The social network, which made the announcement at its annual F8 developers' conference, is launching a series of tutorials and examples to show how machine learning can use multiple GPUs in one machine or be implemented in many machines with one GPU.

It said the aim is to bring machine learning away from its traditional home of data centres and supercomputers to iOS, Android and Raspberry Pi.

It said pre-prepared models from its Caffe2 "Model Zoo" can be run with only a few lines of code.

Facebook, which has been using Caffe2 to train machine learning models, is working with NVIDIA, Qualcomm, Intel, Amazon and Microsoft.

It said: "These collaborations will allow the machine learning community to rapidly experiment using more complex models and deploy the next generation of AI-enhanced apps and services to optimize Caffe2 for both cloud and mobile environments.

"These collaborations will allow the machine learning community to rapidly experiment using more complex models and deploy the next generation of AI-enhanced apps and services."

Qualcomm said it would use its neural processing engine framework in the development of Caffe2. A spokesperson said: "With Caffe2’s modern computation graph design, minimalist modularity, and flexibility to port to multiple platforms, developers can have greater flexibility to design a range of deep learning tasks including computer vision, natural language processing, augmented reality, and event prediction, among others."