Fork me on GitHub
#data-science
<
2019-09-04
>
David Pham20:09:57

I know we should all try to support MXNet and try our project with clojure in it, but this is really cool for CLJS with tensorflow for react native: https://github.com/tensorflow/tfjs/blob/master/tfjs-react-native/README.md

gigasquid20:09:30

Sure - I'm all for that

gigasquid20:09:52

Whatever makes Clojure & Deep Learning stronger

gigasquid20:09:12

Having multiple options is always the best

David Pham20:09:52

Sometimes I feel ML and Clojure would profit from a bit of centralization

David Pham20:09:57

For the initial bit

David Pham20:09:08

Much like ring and the web app world.

gigasquid20:09:43

Well right now it depends on what you want to do - you can't use MXNet for JS only but you can reach into Tensorflow.js

David Pham20:09:24

But you could go with ONNX with MXNet and use ONNX.js

gigasquid20:09:28

You could use tensorflow with clj-python - but then you need python

gigasquid20:09:08

ONNX is a great system, but it doesn't support all models (from what I understand)

David Pham20:09:19

The linked I share make it possible to use TFJS on mobile

gigasquid20:09:25

so your mileage may vary

David Pham20:09:36

Okay I understand.

gigasquid20:09:08

no perfect lunch in any one system yet 🙂

David Pham20:09:46

I wish I could support your on MXNet

David Pham20:09:22

But my job would more likely requires me to go for Nehanderthal/Baydera.

gigasquid20:09:46

Whatever is best for the job 🔧 🙂

David Pham20:09:52

By the way, can you do better flans now? :)

gigasquid20:09:24

Haven't tried lately, but I know more than I did last year - so hopefully 🙂