increase beta

Project/Assets/ML-Agents/Examples/Tennis/Scripts/HitWall.cs

     
     void AgentAWins()
     {
-        m_AgentA.SetReward(.1f);
-        m_AgentB.SetReward(-.1f);// - m_AgentB.timePenalty);
+        m_AgentA.SetReward(1f);
+        m_AgentB.SetReward(-1f);// - m_AgentB.timePenalty);
         m_AgentA.score += 1;
         Reset();
 
     {
-        m_AgentA.SetReward(-.1f);// - m_AgentA.timePenalty);
-        m_AgentB.SetReward(.1f);
+        m_AgentA.SetReward(-1f);// - m_AgentA.timePenalty);
+        m_AgentB.SetReward(1f);
         m_AgentB.score += 1;
         Reset();
 

config/trainer_config.yaml

     batch_size: 2048
     buffer_size: 20480
     hidden_units: 256
-    beta: 1.0e-2
+    beta: 2.0e-2
     time_horizon: 1000
     self_play:
         window: 10