[ConvNets/EffNetV2/TF2] model resume fixed

Author: ntajbakhsh

TensorFlow2/Classification/ConvNets/runtime/runner.py

@@ -282,7 +282,8 @@ def train(self):
 
             target_img_size = self.params.train_img_size
             epochs_per_stage = train_epochs // n_stages 
-            for stage in range(resumed_epoch // epochs_per_stage, n_stages):
+            resumed_stage = min(resumed_epoch // epochs_per_stage, n_stages-1)
+            for stage in range(resumed_stage, n_stages):
                 epoch_st = stage * epochs_per_stage 
                 epoch_end = (epoch_st + epochs_per_stage) if stage < n_stages-1 else train_epochs
                 epoch_curr = epoch_st if epoch_st >= resumed_epoch else resumed_epoch