Increase initialization

ml-agents/mlagents/trainers/ppo/optimizer_torch.py

         value_loss = self.ppo_value_loss(
             values, old_values, returns, decay_eps, loss_masks
         )
-        policy_loss = 2 * self.ppo_policy_loss(
+        policy_loss = self.ppo_policy_loss(
             ModelUtils.list_to_tensor(batch["advantages"]),
             log_probs,
             ModelUtils.list_to_tensor(batch["action_probs"]),

ml-agents/mlagents/trainers/torch/distributions.py

             hidden_size,
             num_outputs,
             kernel_init=Initialization.KaimingHeNormal,
-            kernel_gain=0.1,
+            kernel_gain=0.2,
             bias_init=Initialization.Zero,
         )
         self.tanh_squash = tanh_squash
                 num_outputs,
                 kernel_init=Initialization.KaimingHeNormal,
-                kernel_gain=0.1,
+                kernel_gain=0.2,
                 bias_init=Initialization.Zero,
             )
         else: