how to continue training from a specific checkpoint

[jingang-cv]

Hi,

Thanks for sharing your code.
The training phase needs 1000 epochs.
If I stop the training at the 500 epoch, how to load the model and continue the training?
Just set args.resume=500? Whether need to change the parameter args.start_epoch?

Thank you.

[XuecaiHu]

--pre_train and --start_epoch. We donot use the resume.

[jingang-cv]

Thanks for your reply!