Photos
Image may contain: text that says "1 t:e/0 h:e/0 he:the/0 th:e/0 2 e:the/0 This diagram shows the:the/0 3 simple WFST built in GTN, which transduces word piece decompositions of the" to the word itself."
No photo description available.
Image may contain: text that says "Open Source in Brief FACEBOOK AI"
See all
Videos
With 4.5B parallel sentences in 576 language pairs, CCMatrix is the largest data set of high-quality, web-based bitexts for training translation models. Now Facebook AI is sharing tools for other researchers to use this corpus for their work. https://ai.facebook.com/blog/ccmatrix-a-billion-scale-bitext-data-set-for-training-translation-models/
151
9
With 4.5B parallel sentences in 576 language pairs, CCMatrix is the largest data set of high-quality, web-based bitexts for training translation models. Now Facebook AI is sharing tools for other researchers to use this corpus for their work. https://ai.facebook.com/blog/ccmatrix-a-billion-scale-bitext-data-set-for-training-translation-models/
110
17
We’re releasing fairmotion, a library to help AI researchers use motion capture data for computer graphics and robotics. It provides tools to load, process and visualize motion, and demonstrates its utility with example tasks. We’ve used the library in our work presented at SIGGRAPH 2020 on controlling diverse behaviors for physically simulated characters. Get the code and paper: http://github.com/facebookresearch/fairmotion https://research.fb.com/publications/a-scalable-approach-to-control-diverse-behaviors-for-physically-simulated-characters/
346
16
See all
Posts

Introducing GTN, an open source framework for automatic differentiation with a powerful, expressive type of graph called weighted finite-state transducers (WFSTs). With GTN, AI researchers and engineers can more effectively train such graph-based machine learning models. Read more: https://ai.facebook.com/…/a-new-open-source-framework-for-a…

Image may contain: text that says "1 t:e/0 h:e/0 he:the/0 th:e/0 2 e:the/0 This diagram shows the:the/0 3 simple WFST built in GTN, which transduces word piece decompositions of the" to the word itself."

The future of health care is personal. Today, we’re publishing research on active MRI acquisition, which uses reinforcement learning to adjust AI-accelerated scans on a per-patient basis. This work complements our previous fastMRI research with NYU Langone.

See all