I’m running a simple prototype on a CPU only system. I noticed that you’ve enabled suppression in Docker via torch.backends.nnpack.enabled = False and I’d like to do the same in yolo CLI or Python or both?
It’s a nuisance rather than a problem though.
Yes — this is a PyTorch CPU backend warning rather than an Ultralytics YOLO setting. In Python, set it before creating/running the model:
import torch
torch.backends.nnpack.enabled = False
from ultralytics import YOLO
model = YOLO("yolo26n.pt")
model.predict("image.jpg", device="cpu")
PyTorch also exposes torch.backends.nnpack.flags(enabled=False) for scoped disabling. (docs.pytorch.org)
There isn’t currently a yolo CLI argument for this, so for CLI-style use I’d recommend a tiny wrapper script. Also worth checking with the latest ultralytics and torch, but if inference runs normally this warning is just noise.