The deeplearn script seems broken. It works on the default setting (#define DEEPNET) but when I uncomment #define KERAS and comment the others out and click on Train I get this
How to fix it?

You must install Keras.

https://zorro-project.com/manual/en/deeplearning.htm

It is installed, when its not installed it says init 0 instead of doing this

Ok, next step is finding out what went wrong. Use RStudio to run some keras functions, for instance Model <- keras_model_sequential(). Does it create a model? Can you set up layers?

I get this result.

> Model <- keras_model_sequential()
Loaded Tensorflow version 2.9.1
2022-08-08 13:46:56.508479: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX AVX2
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2022-08-08 13:46:56.955928: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1532] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 5965 MB memory: -> device: 0, name: NVIDIA GeForce RTX 2080, pci bus id: 0000:01:00.0, compute capability: 7.5
> Model
Model: <no summary available, model was not built>
>

I am completely new to R and RStudio so I'm not sure how to setup layers. I come from c# dev

I don't know the error message, but it looks as if Keras was installed for CPU and Tensorflow for GPU, so they don't work together.

For the layers, use this command:

Model %>%
layer_dense(units=30,activation='relu',input_shape = c(ncol(X))) %>%
layer_dropout(rate = 0.2) %>%
layer_dense(units = 1, activation = 'sigmoid')

You can copy it from the keras train function on the manual page, and also the subsequent compile command. But I suppose it won't work when the model was not built. If so, deinstall Keras and Tensorflow and install it again. Make sure that both are installed for CPU, not GPU.

hmm well it does work if I do it like this. Reinstalled completely still the same problem.
can you verify that the keras example works on your machine? (and if it works, which version of python and R?

the RTest works fine btw.

edit: the code from the manual (https://zorro-project.com/manual/en/deeplearning.htm) also doesn't work for me.

All examples work. They are in our test suite that we always run prior to any release. But it's certainly not the newest Keras version.

We'll check in the next days if there are issues with the newest versions of Keras and Tensorflow. Please stay tuned, I'll post here when we find something.

Something I also noticed is that zorro works as 32bit and uses the 64bit exe of RTerm.exe . not sure if that would cause any issues.
Also my gpu did start running and then it supposedly crashes. Out of memory maybe?

A GPU can normally not run out of memory. If it does, it's a hardware issue.

64 bit RTerm is ok and works with all Zorro versions.