Closed zengkaiwei closed 3 years ago
I usually set "--num-steps-stop" for the learning rate scheduler to 150000, and check ckpts between 100000. But it is true that you may also want to check the ckpts around 100000 as there are some randomness for the best ckpts, or you can try multiple runs and check the best score you will get.
When I run train.py with --LB=0.01 as the command you show, I get files about: gta5_2500.pth~gta5_100000.pth, I guess you will test this file and rename the best as gta2city_LB_0_01.pth for multiple test. But I can I find the best model of single beta?