Very cool! Seems like a great way to scale up deployment of inference models. I wonder if this means that they will extend V8 to allow Tensorflow JS to run on a CUDA backend rather than WebGL?

Edit: It seems like TensorflowJS already has a native tensor flow backend! (https://www.tensorflow.org/js/guide/platform_environment#nod...)

AI is amazing. Is very difficult coding for V8, nodejs is very easy for a noob like me.

When support nodejs, because V8 is very limited in library’s.