Thank you :-)

Chainer got my attention with their excellent benchmark (full of details)
https://chainer.org/general/2017/02/08/Performance-of-Distributed-Deep-Learning-Using-ChainerMN.html
and when it appears in @fchollet twitter’s profile
https://twitter.com/fchollet/status/883164419802275840
however, I didn’t tested it yet.

I’ve just had a quick glance and it seems very intuitive to use. And it is very well documented with carefully chosen examples. Doing parallelization is not something trivial, and they have high-level primitives to do so. Definitely, Chainer deserves some attention.

    Ricardo Guerrero Gomez-Ol

    Written by

    Soy científico de datos en BEEVA. Dentro del área de Innovación analizo distintas tecnologías de Deep Learning, Visión Artificial y NLP en combinación con GPUs