In case of pause during learning

AlexeyAB / darknet

YOLOv4 / Scaled-YOLOv4 / YOLO - Neural Networks for Object Detection (Windows and Linux version of Darknet )

http://pjreddie.com/darknet/

Other

21.8k stars 7.97k forks source link

In case of pause during learning #5111

Open Rhee-site opened 4 years ago

Rhee-site commented 4 years ago

Hi, @AlexeyAB

If you are learning using Yolov3.cfg, there will be no problems, but if you are learning using csresnext50-panet-spp.cfg, please contact us as we continue to have problems.

If learning stops due to a problem such as Cuda_Free() or Cuda_Push_array(), please inquire if there is no problem by re-running the training using the last weights file in the backup folder.

We appreciate your answers.

AlexeyAB commented 4 years ago

@Rhee-site Hi,

To catch an error:

show your training command
try to train with random=0, do you get an error?
run training with -cuda_debug_sync -benchmark_layers and show an error message
show screenshot of this info:

sctrueew commented 4 years ago

@Rhee-site Hi,

I have the same problem I train any models without any problem but I get the error when I use csresnext50-panet-spp.cfg or csresnext50-panet-opt.cfg. Did you solve it?

sctrueew commented 4 years ago

@AlexeyAB Hi,

We had a discussion at these issues: 4652 and 5053