Scaling Up (Part 2)
47
Google Developers2.57 млн
Следующее
3 дня – 2007:24Популярные
23 дня – 31 8790:12
25 дней – 5 5321:22Опубликовано 4 декабря 2025, 5:01
Welcome back to part two of our three part series on sharding and parallelism! Let’s explore how to scale your Flax NNX models using JAX's powerful distributed computing capabilities, specifically its SPMD paradigm. If you're coming from PyTorch and have started using JAX and Flax NNX, you know that modern models often outgrow single accelerators. We’ll discuss JAX's approach to parallelism, and how NNX integrates with it seamlessly. This episode focuses on the main workflow: how to integrate JAX's sharding primitives with Flax NNX, with a special focus on the critical sharded initialization pattern.
Resources:
Learn more → goo.gle/learning-jax
Subscribe to Google for Developers → goo.gle/developers
Speaker: Robert Crowe
Resources:
Learn more → goo.gle/learning-jax
Subscribe to Google for Developers → goo.gle/developers
Speaker: Robert Crowe
Свежие видео
4 дня – 4200:22
4 дня – 2 5510:23
5 дней – 5 9681:11
Случайные видео
59 дней – 11 1587:39
126 дней – 15 4042:58
343 дня – 223 80019:04
09.11.24 – 1 4500:59
27.08.24 – 1 9120:42
30.04.16 – 85 6895:15
1 день – 22 8169:40
2 дня – 4 788 18817:25
3 дня – 26 21810:35
4 дня – 5 3760:15
5 дней – 6 0400:47
6 дней – 6551:02
7 дней – 45 93710:53
3 дня – 4810:02