Epoch Starts from 1 When Using --continue_train

[H-skyfxxcker]

Hello everyone,

I have a question regarding the use of the --continue_train flag in my training process. After applying this flag, the terminal shows that the epoch count starts from 1. This has made me uncertain whether I am actually continuing the previous training or if it’s starting over from scratch.

I had previously saved the training state, but with this command, it seems like the training is restarting. Is this normal behavior? Or might I have made an error somewhere? What settings or files should I check to troubleshoot this?

Thank you for your assistance!

[AdairMaldonado]

You can set --epoch_count xx, where xx is your current epoch.