Closed xxyll closed 2 years ago
Hello,
I have never encountered this error, the one at the beginning "OSError: [Errno 24] Too many open files". It doesn't seem to come from my code, so you could try some solutions proposed in this stackoverflow thread : Python Subprocess: Too Many Open Files
Good luck!
Thanks, I solved this problem!!! I'm running 'train' code, it may take a day. I see there is '-pretrained path/to/pretrained_model' in 'test'. I would like to ask in advance, will the 'pretrained_model' be generated after training?
Hello, unfortunately no if you didn't specify the -save_ckpt
parameter.
I will change that so that the default behavior is to save checkpoints. Sorry for the inconvenience.
Oh I see it. I‘m going to retrain and wait for results~
Hello, unfortunately no if you didn't specify the
-save_ckpt
parameter.
I’d like to ask that the -save_ckpt
set up 'save_top_k=3', do I need to adjust the parameters according to the actual epochs?
After training may generate a file named 'ckpt-od-gen1-vgg-11', is that be the path of 'pretrained_model'?
No you don't need to adjust save_top_k, it's the number of checkpoints saved by Pytorch Lightning. 3 means that it will save the 3 best models according to the specified metric. Check out the documentation of Pytorch Lightning for more information. Yes that would be the path of pretrained model.
Hello, Today I try to run
python object_detection.py -path ./PropheseeDataset -backbone vgg-11 -T 5 -tbin 2 -b 8 -epochs 2
. During the first epoch of training, it reports the 'EOFError', do you know how to solve it?