no code implementations • 18 May 2022 • Kapil Vaswani, Stavros Volos, Cédric Fournet, Antonio Nino Diaz, Ken Gordon, Balaji Vembu, Sam Webster, David Chisnall, Saurabh Kulkarni, Graham Cunningham, Richard Osborne, Dan Wilkinson
We present IPU Trusted Extensions (ITX), a set of experimental hardware extensions that enable trusted execution environments in Graphcore's AI accelerators.
no code implementations • • John Winn, Matteo Venanzi, Tom Minka, Ivan Korostelev, John Guiver, Elena Pochernina, Pavel Mishkov, Alex Spengler, Denise Wilkins, Sian Lindley, Richard Banks, Sam Webster, Yordan Zaykov
The knowledge discovery process uses a probabilistic program defining the process of generating the data item from a set of unknown typed entities.
The use of a probabilistic program allows uncertainty in the text to be propagated through to the retrieved facts, which increases accuracy and helps merge facts from multiple sources.
Through an implementation on multi-core CPUs, we show that AMP training converges to the same accuracy as conventional synchronous training algorithms in a similar number of epochs, but utilizes the available hardware more efficiently even for small minibatch sizes, resulting in significantly shorter overall training times.