Spaces:
Running
on
Zero
Running
on
Zero
Solution: torch.OutOfMemoryError: CUDA out of memory: torch.cuda.empty_cache()
#4
by
UniqueFlow
- opened
There's an issue with generating smaller objects once running into cuda out of memory.
Adding "torch.cuda.empty_cache()" to each run in gradio_app e.g. line 348 mitigates this