Model doesnt resume training
syasir1 opened this issue · comments
syasir1 commented
Hey,
I am trying to train the model as per your paper i.e., 300 epochs with 1e-2 lr, and then for 100 epochs with lr 1e-5. but the model dos not resume training even when the checkpoint_path argument is passed. it just runs for one epoch and shows 0.00 accuracy, and ends training. Any suggestion as to how I can resolve said issue?
Regards,
Yasir
Torben Teepe commented
That's how it was intended. The second cycle trains as it's own training.