Rename MADDPG to NDDPG

As the current implementation does not really follow the MADDPG algorithm as described in the paper (link at the bottom), but rather just trains two independent DDPG algorithms, I decided to rename it, to avoid confusion. https://arxiv.org/abs/1706.02275
SwamyDev · Mar 14, 2020 · 824b0ba · 824b0ba
1 parent e66eab6
commit 824b0ba
Show file tree

Hide file tree

Showing 5 changed files with 10 additions and 10 deletions.
diff --git a/resources/models/p3_tennis_final/type.meta b/resources/models/p3_tennis_final/type.meta
diff --git a/tests/test_maddpg_agent_learns.py → tests/test_nddpg_agent_learns.py b/tests/test_maddpg_agent_learns.py → tests/test_nddpg_agent_learns.py
@@ -6,7 +6,7 @@
 
 from tests.auxiliary import GymSession
 from udacity_rl.agents import agent_save, agent_load
-from udacity_rl.agents.maddpg_agent import MADDPGAgent
+from udacity_rl.agents.nddpg_agent import NDDPGAgent
 from udacity_rl.epsilon import NoiseFixed
 
 
@@ -46,7 +46,7 @@ def n_knob():
 @pytest.fixture
 def make_agent(n_knob):
     def factory(**kwargs):
-        return MADDPGAgent(n_knob.observation_space, n_knob.action_space, **kwargs)
+        return NDDPGAgent(n_knob.observation_space, n_knob.action_space, **kwargs)
 
     return factory
 

diff --git a/udacity_rl/agents/__init__.py b/udacity_rl/agents/__init__.py
@@ -4,14 +4,14 @@
 
 from udacity_rl.agents.dqn_agent import DQNAgent
 from udacity_rl.agents.ddpg_agent import DDPGAgent
-from udacity_rl.agents.maddpg_agent import MADDPGAgent
+from udacity_rl.agents.nddpg_agent import NDDPGAgent
 
 logger = logging.getLogger(__name__)
 
 _CLASS_MAPPING = {
     DQNAgent.__name__: DQNAgent,
     DDPGAgent.__name__: DDPGAgent,
-    MADDPGAgent.__name__: MADDPGAgent,
+    NDDPGAgent.__name__: NDDPGAgent,
 }
 
 

diff --git a/udacity_rl/agents/maddpg_agent.py → udacity_rl/agents/nddpg_agent.py b/udacity_rl/agents/maddpg_agent.py → udacity_rl/agents/nddpg_agent.py
@@ -38,7 +38,7 @@
 }
 
 
-class MADDPGAgent(MemoryAgent):
+class NDDPGAgent(MemoryAgent):
     def __init__(self, observation_space, action_space, actor=None, critic=None, **kwargs):
         super().__init__(observation_space, action_space, actor=actor, critic=critic, **kwargs)
 
@@ -62,7 +62,7 @@ def __init__(self, observation_space, action_space, actor=None, critic=None, **k
         self._step = 0
 
     def _print_config(self):  # pragma: no cover
-        logger.info(f"MADDPG configuration:\n"
+        logger.info(f"NDDPG configuration:\n"
                     f"\tNumber of agents:\t{self._num_agents}\n"
                     f"\tObservation Size:\t{self._observation_size}\n"
                     f"\tAction Size:\t\t{self._action_size}\n")

diff --git a/udacity_rl/main.py b/udacity_rl/main.py
@@ -14,7 +14,7 @@
 from udacity_rl.adapter import GymAdapter
 from udacity_rl.agents import DQNAgent, agent_load, agent_save, AgentSnapshot
 from udacity_rl.agents.ddpg_agent import DDPGAgent
-from udacity_rl.agents.maddpg_agent import MADDPGAgent
+from udacity_rl.agents.nddpg_agent import NDDPGAgent
 from udacity_rl.epsilon import EpsilonExpDecay, NoiseFixed
 
 logger = logging.getLogger(__name__)
@@ -41,7 +41,7 @@ class AgentFactory:
     _AGENT_MAPPING = {
         'DQN': DQNAgent,
         'DDPG': DDPGAgent,
-        'MADDPG': MADDPGAgent,
+        'NDDPG': NDDPGAgent,
     }
 
     def __init__(self, algorithm_name):