Add non-deterministic evalation

enjoy.py

@@ -19,8 +19,12 @@
                     help='directory to save agent logs (default: ./trained_models/)')
 parser.add_argument('--add-timestep', action='store_true', default=False,
                     help='add timestep to observations')
+parser.add_argument('--non-det', action='store_true', default=False,
+                    help='whether to use a non-deterministic policy')
 args = parser.parse_args()
 
+args.det = not args.non_det
+
 env = make_vec_envs(args.env_name, args.seed + 1000, 1,
                             None, None, args.add_timestep, device='cpu',
                             allow_early_resets=False)
@@ -56,7 +60,7 @@
 while True:
     with torch.no_grad():
         value, action, _, recurrent_hidden_states = actor_critic.act(
-            obs, recurrent_hidden_states, masks, deterministic=True)
+            obs, recurrent_hidden_states, masks, deterministic=args.det)
 
     # Obser reward and next obs
     obs, reward, done, _ = env.step(action)