The Zorro to R connection is completely different from the Zorro to Python connection. You only need the same bit rate with Python.

I'm running the Zorro 32 to R 64 bit connection for over a year and never had any issues.

Edit: I just noticed JCL's reply

Mind sending me a code example that is verified to work?

Check the RTest script

I did, nothing really stood out. What version of R are you guys using?

It definitely can run out of memory, it can be a configuration issue.

Is there any way to limit max gpu memory usage?

Fixed by downgrading R version used.

There's no need to downgrade R when you adjust the RTermPath setting in Zorro.ini or simply install/update R in a default directory minus the version number, 'C:\Program Files\R\' in my case. Then you only need to adjust the RTermPath path once. Make sure you backup Zorro.ini as it will be overwritten when you install a Zorro update.

I use R 4.2.1

I only use a CPU for processing, so I can't help you with your GPU issue.

I believe the R version does not matter, but a new R installation could solve issue with incompable Keras and Tensorflow versions. We found that when Tensorflow or Anaconda was already installed on a PC, Keras just uses the installed versions regardless whether they match or not. That was probably the reason for the issue.

I'm having the same issue here. it runs the first training sample then fails the second and subsequent ones.

Running training in RStudio works fine with keras and tensorflow with many different model structures.

The GPU is in use for that first training loop that works so I don't think its a tensorflow/keras issue as such, maybe more an issue saving?

Is there a way to make the logs more verbose in terms of any R logs? 7+DIAG only gives the zorro logs.

When you train with the GPU, disable multiple cores. PCs have multiple cores but normally only 1 GPU. There can be also other restrictions with GPU training.