3 ms·

For this to gain traction needs to be adopted by the data-science community. It does not feel like that.

by pantulis 6y ago

sfusato 6y ago
How did you measured that already? The "baby" is yet to be born. We found it's "name" just a few hours ago.
- pantulis 6y ago
  Fair point, let me try and ellaborate. Of course, time will tell. But the data science community fixation is on algorithms, not on languages. So you would need to build Tensorflow or Pytorch upon Nx, and you would need to migrate people from the easy Python/Anaconda/Jupyter to the BEAM world. For that to happen the upside should be obviously evident, I guess in terms of performance. Take, for example, the web world. I believe Phoenix is technically the best framework out there, or at least on par with Rails. And it has not taken the web world by storm (yet). Why would Nx be a success without the support of Google or someone that big?
  - detaro 6y ago
    Why must "take the X world by storm" be the goal?
dnautics 6y ago
I'll tell you why I'm cautiously bullish on Nx. 1) python really does not have broad cultural discipline around testing. Especially not in the ml field. 2) deploying python is not the best. 3) unlike TF, torch is actually not hard to understand and the code should be easily movable and portable out of it. Small time ML users will care about 1&2. 3 makes it relatively painless.
bcardarella 6y ago
The project isn't even public yet, has been under development for only a few months, hasn't had the opportunity to get wider academic eyeballs on it yet, and you're already calling it DOA. Nice.
- m8s 6y ago
  Yeah, this seems like a real disingenuous complaint around a pretty incredible (and still super young) accomplishment. I would also assert that Elixir is an excellent language syntactically to work on data science-y projects. I'm really excited.
Kototama 6y ago
We will see. Can you move local numpys computations on a cluster by just adding a few lines of code? Because this certainly will be possible with Elixir. Most importantly it may allow Elixir developers to not switch to C or Rust for intensive computations (hopefully).

For this to gain traction needs to be adopted by the data-science community. It does not feel like that.

sfusato 6y ago

How did you measured that already? The "baby" is yet to be born. We found it's "name" just a few hours ago.

pantulis 6y ago
Fair point, let me try and ellaborate. Of course, time will tell. But the data science community fixation is on algorithms, not on languages. So you would need to build Tensorflow or Pytorch upon Nx, and you would need to migrate people from the easy Python/Anaconda/Jupyter to the BEAM world. For that to happen the upside should be obviously evident, I guess in terms of performance. Take, for example, the web world. I believe Phoenix is technically the best framework out there, or at least on par with Rails. And it has not taken the web world by storm (yet). Why would Nx be a success without the support of Google or someone that big?
- detaro 6y ago
  Why must "take the X world by storm" be the goal?

dnautics 6y ago

I'll tell you why I'm cautiously bullish on Nx. 1) python really does not have broad cultural discipline around testing. Especially not in the ml field. 2) deploying python is not the best. 3) unlike TF, torch is actually not hard to understand and the code should be easily movable and portable out of it. Small time ML users will care about 1&2. 3 makes it relatively painless.

bcardarella 6y ago

The project isn't even public yet, has been under development for only a few months, hasn't had the opportunity to get wider academic eyeballs on it yet, and you're already calling it DOA. Nice.

m8s 6y ago
Yeah, this seems like a real disingenuous complaint around a pretty incredible (and still super young) accomplishment. I would also assert that Elixir is an excellent language syntactically to work on data science-y projects. I'm really excited.

Kototama 6y ago

We will see. Can you move local numpys computations on a cluster by just adding a few lines of code? Because this certainly will be possible with Elixir. Most importantly it may allow Elixir developers to not switch to C or Rust for intensive computations (hopefully).