Merge pull request #358 from prabhatnagarajan/train_and_eval

Changes variable names in train_agent_with_evaluation

chainerrl/experiments/train_agent.py

@@ -91,10 +91,10 @@ def train_agent(agent, env, steps, outdir, max_episode_len=None,
 def train_agent_with_evaluation(agent,
                                 env,
                                 steps,
-                                eval_n_runs,
+                                eval_n_episodes,
                                 eval_interval,
                                 outdir,
-                                max_episode_len=None,
+                                train_max_episode_len=None,
                                 step_offset=0,
                                 eval_max_episode_len=None,
                                 eval_env=None,
@@ -103,27 +103,27 @@ def train_agent_with_evaluation(agent,
                                 save_best_so_far_agent=True,
                                 logger=None,
                                 ):
-    """Train an agent while regularly evaluating it.
+    """Train an agent while periodically evaluating it.
 
     Args:
-        agent: Agent to train.
-        env: Environment train the againt against.
-        steps (int): Number of total time steps for training.
-        eval_n_runs (int): Number of runs for each time of evaluation.
+        agent: A chainerrl.agent.Agent
+        env: Environment train the agent against.
+        steps (int): Total number of timesteps for training.
+        eval_n_episodes (int): Number of episodes at each evaluation phase
         eval_interval (int): Interval of evaluation.
-        outdir (str): Path to the directory to output things.
-        max_episode_len (int): Maximum episode length.
+        outdir (str): Path to the directory to output data.
+        train_max_episode_len (int): Maximum episode length during training.
         step_offset (int): Time step from which training starts.
         eval_max_episode_len (int or None): Maximum episode length of
-            evaluation runs. If set to None, max_episode_len is used instead.
+            evaluation runs. If None, train_max_episode_len is used instead.
         eval_env: Environment used for evaluation.
         successful_score (float): Finish training if the mean score is greater
-            or equal to this value if not None
+            than or equal to this value if not None
         step_hooks (list): List of callable objects that accepts
             (env, agent, step) as arguments. They are called every step.
             See chainerrl.experiments.hooks.
-        save_best_so_far_agent (bool): If set to True, after each evaluation,
-            if the score (= mean return of evaluation episodes) exceeds
+        save_best_so_far_agent (bool): If set to True, after each evaluation
+            phase, if the score (= mean return of evaluation episodes) exceeds
             the best-so-far score, the current agent is saved.
         logger (logging.Logger): Logger used in this function.
     """
@@ -136,10 +136,10 @@ def train_agent_with_evaluation(agent,
         eval_env = env
 
     if eval_max_episode_len is None:
-        eval_max_episode_len = max_episode_len
+        eval_max_episode_len = train_max_episode_len
 
     evaluator = Evaluator(agent=agent,
-                          n_runs=eval_n_runs,
+                          n_runs=eval_n_episodes,
                           eval_interval=eval_interval, outdir=outdir,
                           max_episode_len=eval_max_episode_len,
                           env=eval_env,
@@ -150,7 +150,7 @@ def train_agent_with_evaluation(agent,
 
     train_agent(
         agent, env, steps, outdir,
-        max_episode_len=max_episode_len,
+        max_episode_len=train_max_episode_len,
         step_offset=step_offset,
         evaluator=evaluator,
         successful_score=successful_score,

examples/ale/train_categorical_dqn_ale.py

@@ -148,10 +148,10 @@ def phi(x):
     else:
         experiments.train_agent_with_evaluation(
             agent=agent, env=env, steps=args.steps,
-            eval_n_runs=args.eval_n_runs, eval_interval=args.eval_interval,
+            eval_n_episodes=args.eval_n_runs, eval_interval=args.eval_interval,
             outdir=args.outdir,
             save_best_so_far_agent=False,
-            max_episode_len=args.max_episode_len,
+            train_max_episode_len=args.max_episode_len,
             eval_env=eval_env,
         )
 

examples/ale/train_dqn_ale.py

@@ -231,10 +231,10 @@ def phi(x):
     else:
         experiments.train_agent_with_evaluation(
             agent=agent, env=env, steps=args.steps,
-            eval_n_runs=args.eval_n_runs, eval_interval=args.eval_interval,
+            eval_n_episodes=args.eval_n_runs, eval_interval=args.eval_interval,
             outdir=args.outdir,
             save_best_so_far_agent=False,
-            max_episode_len=args.max_episode_len,
+            train_max_episode_len=args.max_episode_len,
             eval_env=eval_env,
         )
 

examples/ale/train_ppo_ale.py

@@ -163,9 +163,9 @@ def clip_eps_setter(env, agent, value):
             eval_env=eval_env,
             outdir=args.outdir,
             steps=args.steps,
-            eval_n_runs=args.eval_n_runs,
+            eval_n_episodes=args.eval_n_runs,
             eval_interval=args.eval_interval,
-            max_episode_len=args.max_episode_len,
+            train_max_episode_len=args.max_episode_len,
             save_best_so_far_agent=False,
             step_hooks=[
                 lr_decay_hook,

examples/atari/dqn/train_dqn.py

@@ -174,10 +174,10 @@ def phi(x):
     else:
         experiments.train_agent_with_evaluation(
             agent=agent, env=env, steps=args.steps,
-            eval_n_runs=args.eval_n_runs, eval_interval=args.eval_interval,
+            eval_n_episodes=args.eval_n_runs, eval_interval=args.eval_interval,
             outdir=args.outdir,
             save_best_so_far_agent=False,
-            max_episode_len=args.max_episode_len,
+            train_max_episode_len=args.max_episode_len,
             eval_env=eval_env,
         )
 

examples/gym/train_categorical_dqn_gym.py

@@ -163,9 +163,9 @@ def make_env(test):
     else:
         experiments.train_agent_with_evaluation(
             agent=agent, env=env, steps=args.steps,
-            eval_n_runs=args.eval_n_runs, eval_interval=args.eval_interval,
+            eval_n_episodes=args.eval_n_runs, eval_interval=args.eval_interval,
             outdir=args.outdir, eval_env=eval_env,
-            max_episode_len=timestep_limit)
+            train_max_episode_len=timestep_limit)
 
 
 if __name__ == '__main__':

examples/gym/train_ddpg_gym.py

@@ -171,9 +171,9 @@ def random_action():
         experiments.train_agent_with_evaluation(
             agent=agent, env=env, steps=args.steps,
             eval_env=eval_env,
-            eval_n_runs=args.eval_n_runs, eval_interval=args.eval_interval,
+            eval_n_episodes=args.eval_n_runs, eval_interval=args.eval_interval,
             outdir=args.outdir,
-            max_episode_len=timestep_limit)
+            train_max_episode_len=timestep_limit)
 
 
 if __name__ == '__main__':

examples/gym/train_dqn_gym.py

@@ -197,9 +197,9 @@ def make_env(test):
     else:
         experiments.train_agent_with_evaluation(
             agent=agent, env=env, steps=args.steps,
-            eval_n_runs=args.eval_n_runs, eval_interval=args.eval_interval,
+            eval_n_episodes=args.eval_n_runs, eval_interval=args.eval_interval,
             outdir=args.outdir, eval_env=eval_env,
-            max_episode_len=timestep_limit)
+            train_max_episode_len=timestep_limit)
 
 
 if __name__ == '__main__':

examples/gym/train_pcl_gym.py

@@ -223,9 +223,9 @@ def make_env(process_idx, test):
                 eval_env=make_env(0, test=True),
                 outdir=args.outdir,
                 steps=args.steps,
-                eval_n_runs=args.eval_n_runs,
+                eval_n_episodes=args.eval_n_runs,
                 eval_interval=args.eval_interval,
-                max_episode_len=timestep_limit)
+                train_max_episode_len=timestep_limit)
 
 
 if __name__ == '__main__':

examples/gym/train_ppo_gym.py

@@ -203,9 +203,9 @@ def clip_eps_setter(env, agent, value):
             eval_env=make_env(True),
             outdir=args.outdir,
             steps=args.steps,
-            eval_n_runs=args.eval_n_runs,
+            eval_n_episodes=args.eval_n_runs,
             eval_interval=args.eval_interval,
-            max_episode_len=timestep_limit,
+            train_max_episode_len=timestep_limit,
             save_best_so_far_agent=False,
             step_hooks=[
                 lr_decay_hook,

examples/gym/train_reinforce_gym.py

@@ -140,9 +140,9 @@ def make_env(test):
             eval_env=eval_env,
             outdir=args.outdir,
             steps=args.steps,
-            eval_n_runs=args.eval_n_runs,
+            eval_n_episodes=args.eval_n_runs,
             eval_interval=args.eval_interval,
-            max_episode_len=timestep_limit)
+            train_max_episode_len=timestep_limit)
 
 
 if __name__ == '__main__':

examples/gym/train_trpo_gym.py

@@ -200,9 +200,9 @@ def make_env(test):
             eval_env=make_env(test=True),
             outdir=args.outdir,
             steps=args.steps,
-            eval_n_runs=args.eval_n_runs,
+            eval_n_episodes=args.eval_n_runs,
             eval_interval=args.eval_interval,
-            max_episode_len=timestep_limit,
+            train_max_episode_len=timestep_limit,
         )
 
 

examples/quickstart/quickstart.ipynb

@@ -465,8 +465,8 @@
     "chainerrl.experiments.train_agent_with_evaluation(\n",
     "    agent, env,\n",
     "    steps=2000,           # Train the agent for 2000 steps\n",
-    "    eval_n_runs=10,       # 10 episodes are sampled for each evaluation\n",
-    "    max_episode_len=200,  # Maximum length of each episodes\n",
+    "    eval_n_episodes=10,       # 10 episodes are sampled for each evaluation\n",
+    "    train_max_episode_len=200,  # Maximum length of each episode\n",
     "    eval_interval=1000,   # Evaluate the agent after every 1000 steps\n",
     "    outdir='result')      # Save everything to 'result' directory"
    ]

tests/agents_tests/basetest_training.py

@@ -52,7 +52,7 @@ def _test_training(self, gpu, steps=5000, load_model=False,
         # Train
         train_agent_with_evaluation(
             agent=agent, env=env, steps=steps, outdir=self.tmpdir,
-            eval_interval=200, eval_n_runs=5, successful_score=1,
+            eval_interval=200, eval_n_episodes=5, successful_score=1,
             eval_env=test_env)
 
         agent.stop_episode()

tests/agents_tests/test_pcl.py

@@ -215,9 +215,9 @@ def phi(x):
                 eval_env=make_env(0, True),
                 outdir=self.outdir,
                 steps=steps,
-                max_episode_len=2,
+                train_max_episode_len=2,
                 eval_interval=200,
-                eval_n_runs=5,
+                eval_n_episodes=5,
                 successful_score=1)
 
         agent.stop_episode()

tests/agents_tests/test_ppo.py

@@ -79,10 +79,10 @@ def _test_abc(self, steps=1000000,
             steps=steps,
             outdir=self.tmpdir,
             eval_interval=200,
-            eval_n_runs=50,
+            eval_n_episodes=50,
             successful_score=successful_return,
             eval_env=test_env,
-            max_episode_len=max_episode_len,
+            train_max_episode_len=max_episode_len,
         )
 
         agent.stop_episode()

tests/agents_tests/test_reinforce.py

@@ -137,9 +137,9 @@ def phi(x):
             eval_env=make_env(0, True),
             outdir=self.outdir,
             steps=steps,
-            max_episode_len=2,
+            train_max_episode_len=2,
             eval_interval=500,
-            eval_n_runs=5,
+            eval_n_episodes=5,
             successful_score=1)
 
         # Test

tests/agents_tests/test_trpo.py

@@ -203,9 +203,9 @@ def _test_abc(self, steps=1000000,
             steps=steps,
             outdir=self.tmpdir,
             eval_interval=200,
-            eval_n_runs=5,
+            eval_n_episodes=5,
             successful_score=successful_return,
-            max_episode_len=max_episode_len,
+            train_max_episode_len=max_episode_len,
         )
 
         agent.stop_episode()