Added testing during training to evaluate the current training status

environment.py

@@ -20,6 +20,9 @@ def get_env(self):
         """
         return self._env, self._params
 
+    def close_env(self):
+        self._env.close()
+
     def get_goal(self):
         return
 

main.py

@@ -10,7 +10,11 @@
 experiment = "FetchReach-v1"
 env = gym.make(experiment)
 
-# Hyperparameters
+# Program hyperparameters
+TESTING_INTERVAL = 50 # number of updates between two evaluation of the policy
+TESTING_ROLLOUTS = 100 # number of rollouts performed to evaluate the current policy
+
+# Algorithm hyperparameters
 BATCH_SIZE = 32
 BUFFER_SIZE = 100000
 MAX_STEPS = 50 # WARNING: defined in multiple files...
@@ -26,14 +30,16 @@
 replay_buffer = ReplayBuffer(BUFFER_SIZE)
 
 # should be done per episode
-randomized_environment.sample_env()
-env, env_params = randomized_environment.get_env()
 
-success = 0
 
 for ep in range(EPISODES):
     # generate a rollout
 
+    # generate an environment
+    randomized_environment.sample_env()
+    env, env_params = randomized_environment.get_env()
+
+
     # reset the environment
     current_obs_dict = env.reset()
 
@@ -53,7 +59,6 @@
 
     # rollout the  whole episode
     while not done:
-        env.render()
         obs = current_obs_dict['observation']
         history = episode.get_history()
 
@@ -109,9 +114,6 @@
             # WARNING FIXME: needs padding
             t_batch += episodes[i].get_terminal()[1:]
 
-
-        #s_batch, a_batch, r_batch, t_batch, s2_batch, history_batch, env_batch, goal_batch = replay_buffer.sample_batch(BATCH_SIZE)
-
         target_action_batch = agent.evaluate_actor_batch(agent._actor.predict_target, next_s_batch, goal_batch, history_batch)
 
         predicted_actions = agent.evaluate_actor_batch(agent._actor.predict, next_s_batch, goal_batch, history_batch)
@@ -136,3 +138,52 @@
         # Update target networks
         agent.update_target_actor()
         agent.update_target_critic()
+
+        randomized_environment.close_env()
+
+        # perform policy evaluation
+        if ep % TESTING_INTERVAL == 0:
+            success_number = 0
+
+            for test_ep in range(TESTING_ROLLOUTS):
+                randomized_environment.sample_env()
+                env, env_params = randomized_environment.get_env()
+
+                current_obs_dict = env.reset()
+
+                # read the current goal, and initialize the episode
+                goal = current_obs_dict['desired_goal']
+                episode = Episode(goal, env_params, MAX_STEPS)
+
+                # get the first observation and first fake "old-action"
+                # TODO: decide if this fake action should be zero or random
+                obs = current_obs_dict['observation']
+                last_action = env.action_space.sample()
+
+                episode.add_step(last_action, obs, 0)
+
+                done = False
+
+                # rollout the whole episode
+                while not done:
+                    obs = current_obs_dict['observation']
+                    history = episode.get_history()
+
+                    action = agent.evaluate_actor(agent._actor.predict_target, obs, goal, history)
+
+                    new_obs_dict, step_reward, done, info = env.step(action[0])
+
+                    new_obs = new_obs_dict['observation']
+
+                    episode.add_step(action[0], new_obs, step_reward)
+
+                    total_reward += step_reward
+
+                    current_obs_dict = new_obs_dict
+
+                if info['is_success'] > 0.0:
+                    success_number += 1
+
+                randomized_environment.close_env()
+
+            print("Testing at episode {}, success rate : {}".format(ep, success_number/TESTING_ROLLOUTS))