When memory errors occur with model.fit(), is it due to GPU RAM or normal RAM?

Question

With respect to this question,

https://stackoverflow.com/questions/51895278/how-to-know-when-to-use-fit-generator-in-keras-when-training-data-gets-too-big

when memory errors are reported due to excessively large training data, are the memory errors caused by lack of normal PC RAM or lack of GPU RAM? I would like to know which one so that I can buy the right kind of RAM.

Keras was configured to use GPU for training.

Answer 1

Yes. The error is for GPU memory. You should look at training in batches options if you haven't already. This thread has short explanation on possible benefits and implications.

Quick batch size tutorial