add the missing .shape to q value networks

Summary: Add the missing .shape to q value networks.

Reviewed By: rodrigodesalvobraz

Differential Revision: D67272397

fbshipit-source-id: 88cf32b9efa43fd9051c8de17fcc82ec3214fbe7

Author: yiwan-rl

pearl/neural_networks/sequential_decision_making/q_value_networks.py

@@ -172,7 +172,7 @@ def get_q_values(
         q_values = self.forward(x).squeeze(
             -1
         )  # (batch_size, number_of_actions_to_query)
-        return q_values if len(action_batch) == 3 else q_values.squeeze(-1)
+        return q_values if len(action_batch.shape) == 3 else q_values.squeeze(-1)
 
     @property
     def state_dim(self) -> int:
@@ -239,7 +239,7 @@ def get_q_values(
             q_values,  # (batch_size x num actions x 1)
         )  # (batch_size x number of query actions x 1)
         q_values = q_values.squeeze(-1)  # (batch_size x number of query actions)
-        return q_values if len(action_batch) == 3 else q_values.squeeze(-1)
+        return q_values if len(action_batch.shape) == 3 else q_values.squeeze(-1)
 
     @property
     def state_dim(self) -> int:
@@ -324,7 +324,7 @@ def get_q_value_distribution(
         q_values = self.forward(
             x
         )  # (batch_size, number_of_actions_to_query, number_of_quantiles)
-        return q_values if len(action_batch) == 3 else q_values.squeeze(-2)
+        return q_values if len(action_batch.shape) == 3 else q_values.squeeze(-2)
 
     @property
     def quantiles(self) -> Tensor:
@@ -503,7 +503,7 @@ def get_q_values(
             state_value + advantage - advantage_mean
         )  # shape: (batch_size, number of query actions)
 
-        return q_values if len(action_batch) == 3 else q_values.squeeze(-1)
+        return q_values if len(action_batch.shape) == 3 else q_values.squeeze(-1)
 
 
 """
@@ -590,7 +590,7 @@ def get_q_values(
         q_values = self._interaction_features.forward(x).squeeze(
             -1
         )  # (batch_size, number_of_actions_to_query)
-        return q_values if len(action_batch) == 3 else q_values.squeeze(-1)
+        return q_values if len(action_batch.shape) == 3 else q_values.squeeze(-1)
 
     @property
     def state_dim(self) -> int:
@@ -695,7 +695,7 @@ def get_q_values(
         q_values = self.forward(x, z=z, persistent=persistent).squeeze(
             -1
         )  # (batch_size, number_of_actions_to_query)
-        return q_values if len(action_batch) == 3 else q_values.squeeze(-1)
+        return q_values if len(action_batch.shape) == 3 else q_values.squeeze(-1)
 
     @property
     def state_dim(self) -> int:
@@ -806,7 +806,7 @@ def get_q_values(
         q_values = self._model_fc(x).reshape(
             batch_size, num_query_actions
         )  # (batch_size, number_of_actions_to_query)
-        return q_values if len(action_batch) == 3 else q_values.squeeze(-1)
+        return q_values if len(action_batch.shape) == 3 else q_values.squeeze(-1)
 
     @property
     def state_dim(self) -> int: