Alvaro de Castro

Thanks, you are right and actually, that´s what we did: we just let it run for some days and now we are down to 1.45%. We froze the model and...

Thank you: we let it run for some more and it would ping pong between 1 and 2, so since it seems to work well we consider it trained.

I did that and it is still using the CPU: when I put "watch nvidia-smi" it shows 0% gpu usage

Thanks, I do have tensorflow-gpu installed and I´ve used in other projects. It´s enough to open another ssh screen and type "watch nvidia-smi" to see it running. Besides, it is...

After trying with many combinations, the one that worked is tensorflow-gpu==1.13.1 and keras==2.2.4 running under python 2 and tensorflow 10.0.0 Any other combination and you run into all sort of...

ok solved it: it was that it cannot reroute to another site at nginx

Those models say "For now, a 8x80G multi-GPU machine is required", so I fail to see what's the benefit

"For now, a 8x80G multi-GPU machine is required)."

I've seen that error before and it was cured by trying other versions of numpy believe it or not

The problem is not during INSTALLATION, is during EXECUTION As soon as it loads ultralytics, it starts to change torch and torchvision, among other things And in jetson torchvision has...