No video :(

Ray: A System for Scalable Python and ML |SciPy 2020| Robert Nishihara

Enthought

Подписаться 67 тыс.

Просмотров 10 тыс.

50% 1

Видео Поделиться Скачать Добавить в

Опубликовано:

29 авг 2024

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист

Посмотреть позже

Комментарии : 14

@alonhazan6324 4 года назад

I Love Ray. Parallelization that works!

@mohammedal-andoli6458 4 года назад

I'm beginner in ray, Can I use ray on windows or we need WSL2. Also what is the easiest way to build cluster with windows if possiple.

@alonhazan6324 4 года назад

Sorry I use Linux for development. But I hear ray supposed to support windows now.

@kursatsahin74 3 года назад

@@alonhazan6324 Hi Alon. Can you share your configuration and a sample code that works on Ray Distributed with me ??. In our side, we have been testing some codes on Ray 3-Node ( not scaleable one) and python codes with pandas not working properly on Ray Distributed system. I appreciate your comment.

@argmax11 4 года назад

Really great presentation !

@dhirsun 3 года назад

Really simple and great presentation

@dawidlaszuk 4 года назад

What's the AWS cluster that you set up? EC2 instances? ECS? 30 seconds for provisioning 10 units (hosts?) is really good. Surprisingly good.

@user-jx6ys4gn4r 3 года назад

such a nice presentation!

@magedhelmy8512 3 года назад

Amazing! Keep the good work up.

@arshadjaveed7877 2 года назад

Awesome

@slavicasubic2454 4 года назад

Ray rocks

@SrEngr 3 года назад

Chart showing scaling around 12:20 might have done better to have y on a log-10 scale.

@TernaryM01 3 года назад

Yep. It looks like exponential growth, but that's impossible. The best speedup one can achieve by parallelization is linear (which is indeed roughly the case in the chart).