search

EagerAI / tfaddons

R interface to TensorFlow 2.x SIG-Addons
https://eagerai.github.io/tfaddons/
Apache License 2.0
20 stars 0 forks source link

tensorflow-addons-0.14.0 or higher? #10

Open cmclean5 opened 2 years ago

cmclean5 commented 2 years ago

Dear Reader,

I have tensorflow version 2.6 running on a M1 mac, and have been using it through R + Keras API in an analysis. I would also like to use the tensorflow-addons package, however I don't think I can use the given version, i.e. 0.10.0, in my analysis? I use wonder if there is a way to use the tensorflow-addons-0.10.0 package in my current setup, or do I have to down grade my tensorflow version to 2.2?

Kindest Regards, Colin.

cmclean5 commented 2 years ago

Cool, so it works with the setup I have. Running the example on the README page I found (with my setup) an issue using optimizer_radam()...

"" Error in py_call_impl(callable, dots$args, dots$keywords) : TypeError: type of argument "total_steps" must be int; got float instead "" ... I'm not passing the parameter "total_steps" a value, so might be internal? But I got the full example working by just using optimizer_adam()

Cheers, Colin.

turgut090 commented 2 years ago

Hi could you pass this argument manually and set it as integer, please? E.g. total_steps = as.integer(0) If it works, I will fix that with PR

cmclean5 commented 2 years ago

Hiya,

Yeah the example given in the README now works, for my setup away, when specifying total_steps = as.integer(0) in function optimizer_radam.