Merge branch 'integrate-buffer-fix' into develop

pom.xml

@@ -8,7 +8,7 @@
 
   <groupId>de.monticore.lang.monticar</groupId>
   <artifactId>embedded-montiarc-emadl-generator</artifactId>
-  <version>0.3.4-SNAPSHOT</version>
+  <version>0.3.5-SNAPSHOT</version>
 
   <!-- == PROJECT DEPENDENCIES ============================================= -->
 
@@ -20,7 +20,7 @@
     <cnnarch-generator.version>0.0.2-SNAPSHOT</cnnarch-generator.version>
     <cnnarch-mxnet-generator.version>0.2.16-SNAPSHOT</cnnarch-mxnet-generator.version>
     <cnnarch-caffe2-generator.version>0.2.12-SNAPSHOT</cnnarch-caffe2-generator.version>
-    <cnnarch-gluon-generator.version>0.2.5-SNAPSHOT</cnnarch-gluon-generator.version>
+    <cnnarch-gluon-generator.version>0.2.6-SNAPSHOT</cnnarch-gluon-generator.version>
     <embedded-montiarc-math-opt-generator>0.1.4</embedded-montiarc-math-opt-generator>
 
         <!-- .. Libraries .................................................. -->

src/test/resources/target_code/gluon/reinforcementModel/cartpole/CNNTrainer_cartpole_master_dqn.py

@@ -56,7 +56,7 @@ if __name__ == "__main__":
             'memory_size': 10000,
             'sample_size': 32,
             'state_dtype': 'float32',
-            'action_dtype': 'float32',
+            'action_dtype': 'uint8',
             'rewards_dtype': 'float32'
         },
         'strategy_params': {

src/test/resources/target_code/gluon/reinforcementModel/cartpole/reinforcement_learning/agent.py

@@ -1043,6 +1043,14 @@ class DqnAgent(Agent):
         self._double_dqn = double_dqn
         self._use_fix_target = use_fix_target
 
+        # Build memory buffer for discrete actions
+        replay_memory_params['state_dim'] = state_dim
+        replay_memory_params['action_dim'] = (1,)
+        self._replay_memory_params = replay_memory_params
+        rm_builder = ReplayMemoryBuilder()
+        self._memory = rm_builder.build_by_params(**replay_memory_params)
+        self._minibatch_size = self._memory.sample_size
+
         # Initialize best network
         self._best_net = copy_net(self._qnet, self._state_dim, self._ctx)
         self._best_avg_score = -np.infty
@@ -1199,7 +1207,7 @@ class DqnAgent(Agent):
 
                 # 3. Store transition in replay memory
                 self._memory.append(
-                    state, action, reward, next_state, terminal)
+                    state, [action], reward, next_state, terminal)
 
                 # 4. Train the network if in interval
                 if self._do_training():

src/test/resources/target_code/gluon/reinforcementModel/mountaincar/reinforcement_learning/agent.py

@@ -1043,6 +1043,14 @@ class DqnAgent(Agent):
         self._double_dqn = double_dqn
         self._use_fix_target = use_fix_target
 
+        # Build memory buffer for discrete actions
+        replay_memory_params['state_dim'] = state_dim
+        replay_memory_params['action_dim'] = (1,)
+        self._replay_memory_params = replay_memory_params
+        rm_builder = ReplayMemoryBuilder()
+        self._memory = rm_builder.build_by_params(**replay_memory_params)
+        self._minibatch_size = self._memory.sample_size
+
         # Initialize best network
         self._best_net = copy_net(self._qnet, self._state_dim, self._ctx)
         self._best_avg_score = -np.infty
@@ -1199,7 +1207,7 @@ class DqnAgent(Agent):
 
                 # 3. Store transition in replay memory
                 self._memory.append(
-                    state, action, reward, next_state, terminal)
+                    state, [action], reward, next_state, terminal)
 
                 # 4. Train the network if in interval
                 if self._do_training():

src/test/resources/target_code/gluon/reinforcementModel/torcs/CNNTrainer_torcs_agent_torcsAgent_dqn.py

@@ -63,7 +63,7 @@ if __name__ == "__main__":
             'memory_size': 1000000,
             'sample_size': 32,
             'state_dtype': 'float32',
-            'action_dtype': 'float32',
+            'action_dtype': 'uint8',
             'rewards_dtype': 'float32'
         },
         'strategy_params': {

src/test/resources/target_code/gluon/reinforcementModel/torcs/reinforcement_learning/agent.py

@@ -1043,6 +1043,14 @@ class DqnAgent(Agent):
         self._double_dqn = double_dqn
         self._use_fix_target = use_fix_target
 
+        # Build memory buffer for discrete actions
+        replay_memory_params['state_dim'] = state_dim
+        replay_memory_params['action_dim'] = (1,)
+        self._replay_memory_params = replay_memory_params
+        rm_builder = ReplayMemoryBuilder()
+        self._memory = rm_builder.build_by_params(**replay_memory_params)
+        self._minibatch_size = self._memory.sample_size
+
         # Initialize best network
         self._best_net = copy_net(self._qnet, self._state_dim, self._ctx)
         self._best_avg_score = -np.infty
@@ -1199,7 +1207,7 @@ class DqnAgent(Agent):
 
                 # 3. Store transition in replay memory
                 self._memory.append(
-                    state, action, reward, next_state, terminal)
+                    state, [action], reward, next_state, terminal)
 
                 # 4. Train the network if in interval
                 if self._do_training():