We call check_logdir in run.py on line 35, and then again if we are running infer then we call check_logdir again inside create_model inside utils.py on line 764.
To further complicate matters, we sometimes check two different logdirs.
The second check_logdir has been removed and we simply pass the checkpoint recovered from the first check_logdir call to create_model as a parameter as opposed to fetching it again.
We call check_logdir in run.py on line 35, and then again if we are running infer then we call check_logdir again inside create_model inside utils.py on line 764. To further complicate matters, we sometimes check two different logdirs. The second check_logdir has been removed and we simply pass the checkpoint recovered from the first check_logdir call to create_model as a parameter as opposed to fetching it again.
Rebase from #279