TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Using JAX in 2022

66 pointsby SleekEagleover 3 years ago

6 comments

brileeover 3 years ago
This is a really great overview of JAX - the best I've seen outside of primary Google/DeepMind sources. Glad to see people besides Googlers getting familiar with its capabilities.
评论 #30352181 未加载
time_to_smileover 3 years ago
&gt; Why Should I Care About JAX?<p>&gt; In short - speed.<p>For me personally, the magic of JAX is that it able to have this performance, while being as close as possible to having <i>first class differentiation</i> in Python. The latter is a far more important reason to use JAX. It can really change how you think about programming and ML. Rather than implementing a specific model, you can write up the parameterized solution to a problem then <i>solve it</i>.<p>However first class differentiation ultimately isn&#x27;t really useful unless you happen to also solve the speed problem. That is what makes JAX incredible. From the programming perspective JAX is to differentiable programming what Prolog is to logic programming, however Prolog has always been limited ultimately by performance problems where JAX is not.
评论 #30352731 未加载
评论 #30352730 未加载
skrebbelover 3 years ago
With the year so prominently in the title, i thought this was going to be about a technology that is obviously not trendy anymore, ie, JAX The Terrible Java XML Parser. I&#x27;m disappointed that it was just blog title spam about some non-controversial modern technology.
评论 #30351676 未加载
评论 #30362172 未加载
paperworkover 3 years ago
I wish JAX worked with windows natively (without using wsl). I teach a very high level intro to numpy and would _love_ to have my students try jax. These students are relatively new to programming and the idea of using a linux shell or having to compile anything themselves just wouldn&#x27;t work.
评论 #30355436 未加载
评论 #30352691 未加载
azinman2over 3 years ago
These benchmarks are pretty crazy, especially as I presumed NumPy to do far better. Isn’t tensorflow etc already doing GPU acceleration? Is this actually a fair comparison?
评论 #30351773 未加载
评论 #30352025 未加载
评论 #30352029 未加载
评论 #30351972 未加载
评论 #30352050 未加载
评论 #30351957 未加载
otsalomaover 3 years ago
How does JAX&#x27;s JIT compare to Numba?
评论 #30355090 未加载
评论 #30352130 未加载