hyptest

config/ppo/WalkerDynamic.yaml

   WalkerDynamic:
     trainer: ppo
     batch_size: 2048
-    beta: 0.005
+    beta: 0.01
-    epsilon: 0.2
+    epsilon: 0.3
     hidden_units: 512
     lambd: 0.95
     learning_rate: 0.0003
     normalize: true
-    num_epoch: 10
+    num_epoch: 10 #3
     num_layers: 3
     time_horizon: 1000
     sequence_length: 64