Out of memory

I have a question about how to free memory when integrating weights and biases with yolox implementation?
After using weights and biases, my memory crashes before even completing the first epoch
I am using kaggle for yolox training
Any suggestions for this issue?

Is it crashing without W&B?
Using W&B shouldn’t add much memory overhead so it would be pretty unexpected if just adding W&B gave you a OOM error. Typical ways to reduce memory usage is to reduce your batch size, and make sure you’re not accidentally keeping data on the GPU.

Yes sir.
It only crashes only after adding wandb to the code

Sorry for the late response. Can you share some code so we can try reproduce please?

Hi there, I wanted to follow up on this request. Please let us know if we can be of further assistance or if your issue has been resolved.

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.