Just like the title, I can not find the log file when I have trained my own model by my custom data.
Where can I find it or what should I do to create the log file?
I add the following line in logging.py after line 79:
fileNow = time.strftime('%Y-%m-%d',time.localtime(time.time())) + 'train-log.log'
logging.basicConfig(filename = fileNow,level=logging.INFO, format=FORMAT, stream=sys.stdout)
It will creat a log file in $detectron$ recording some log. But the terminal doesn't show training message after that. Anyone has a better idea?
I just added the tee command to start the training:
python2 tools/train_net.py <CONFIGURATIONS> | tee /path/to/log/file.txt
That might not be the best solution but It works. I just use that to visualize the training process. Be aware that I made the experience that errors will not be printed into the log file.
I use nohup python *py > mylog.log & to redirect the outputs to the log file.
Hi @wkoa, logs are written to standard output (stdout) so the logs won't be saved to a regular file by default. See the suggestions above for how you may go about redirecting and saving the logging outputs.
I'm able to log the training since I manipulated the attached files.
The log-file is written to the same folder as model_final.pkl and includes all (so far known) parameters I need to plot a loss-epoch-curve.
Maybe it's helpful for someone :)
can you tell me how to draw the PR curve?
Most helpful comment
I'm able to log the training since I manipulated the attached files.
The log-file is written to the same folder as model_final.pkl and includes all (so far known) parameters I need to plot a loss-epoch-curve.
Maybe it's helpful for someone :)
log_changes.zip