Image Classification From Scratch: Make note of large batch size

[jxu]

Issue Type

Documentation Bug

Source

binary

Keras Version

3.6.0

Custom Code

No

OS Platform and Distribution

Ubuntu 22.04

Python version

3.10

GPU model and memory

Quadro M1200 4 GB

Current Behavior?

Running the example I kept getting OOM trying to allocate 10 GB. Then I remembered my old GPU only has 4 GB memory. Changing batch-size from 128 to 32 fixed it. That might sound obvious, but I after spending half a day trying to install CUDA, I almost gave up and thought it was unsolvable when I encountered the error.

Anyone running a slightly older GPU like the RTX 2080 8 GB will probably get OOM too.

Standalone code to reproduce the issue or tutorial link

https://keras.io/examples/vision/image_classification_from_scratch/

Original error: RESOURCE_EXHAUSTED: Out of memory while trying to allocate 1078460416 bytes.

Relevant log output

No response