diff --git a/MetaAugment/autoaugment_learners/gru_learner.py b/MetaAugment/autoaugment_learners/gru_learner.py
index 5f8e19bef5308a40fa8e51fef2afb7b382f305d8..c06edec316eed6982272abc685d6e02735e92adf 100644
--- a/MetaAugment/autoaugment_learners/gru_learner.py
+++ b/MetaAugment/autoaugment_learners/gru_learner.py
@@ -61,7 +61,7 @@ class gru_learner(aa_learner):
                 # GRU-specific attributes that aren't in all other aa_learners's
                 alpha=0.2,
                 cont_mb_size=4,
-                cont_lr=0.1):
+                cont_lr=0.03):
         """
         Args:
             alpha (float, optional): Exploration parameter. It is multiplied to 
diff --git a/MetaAugment/main.py b/MetaAugment/main.py
index 61dec50cc98134d6931d721c79b2d53fd11c750f..5796c94cb349808ad92f028e2c9318ac3fb27316 100644
--- a/MetaAugment/main.py
+++ b/MetaAugment/main.py
@@ -51,6 +51,7 @@ def train_child_network(child_network,
         device = torch.device('cpu')
     child_network = child_network.to(device=device)
     
+    total_val=torch.tensor([0.0])
     best_acc=torch.tensor([0.0])
     early_stop_cnt = 0