Running a basic model? An MNIST example?

[Jonathan]

Apologies if I say something dumb, I'm cautiously exploring the space right now.

How easily, given a basic pre-trained model 'tf_model.pb' (e.g. for classification of MNIST) could it be used for prediction?

Something similar to: https://www.dlology.com/blog/how-to-convert-trained-keras-model-to-tensorflow-and-make-prediction/ (just the 'Load .pb file and make predictions' section)

I've looked at https://github.com/tensorflow/rust/tree/master/examples but I can't quite fit the parts together, does anyone have an even simpler example like I mentioned (running a pre-trained model, for an mnist-like task)?

I'm exploring options for a project that does some nn based classifcation, if I can grasp this, implementing it with actix seems a decent option.

Sorry for a bit of a stackoverflow-like question, I'm just a bit unsure right now.

[Kylekosic]

Hey Jonathan,

I have an example that sounds quite similar to what you are looking for. Here is a repo I have that loads a .pb mnist model trained in python for the purposes of serving using Actix-web. Hope it helps: 

https://github.com/kykosic/actix-tensorflow-example

On Mar 20, 2021, at 21:58, Jonathan <jonathanwo...@gmail.com> wrote:

Apologies if I say something dumb, I'm cautiously exploring the space right now.

--
You received this message because you are subscribed to the Google Groups "Rust for TensorFlow" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rust+uns...@tensorflow.org.
To view this discussion on the web visit https://groups.google.com/a/tensorflow.org/d/msgid/rust/55fd8103-96a7-4958-be8d-a5f63da2f73en%40tensorflow.org.

[Jonathan]

Looks great, thankyou!