diff --git a/mmaction/models/recognizers/base.py b/mmaction/models/recognizers/base.py
index 954c5bf69fb7fb77e7b886000f96ff6398d57b87..91280716320ce29a2fe831c1d19f0cbaca506100 100644
--- a/mmaction/models/recognizers/base.py
+++ b/mmaction/models/recognizers/base.py
@@ -81,10 +81,13 @@ class BaseRecognizer(nn.Module, metaclass=ABCMeta):
 
     @abstractmethod
     def forward_train(self, imgs, labels):
+        """Defines the computation performed at every call when training."""
         pass
 
     @abstractmethod
     def forward_test(self, imgs):
+        """Defines the computation performed at every call
+        when evaluation and testing."""
         pass
 
     @staticmethod
diff --git a/mmaction/models/recognizers/recognizer2d.py b/mmaction/models/recognizers/recognizer2d.py
index 8dcca86b0a71b993f683028293472f4c15437f5a..e6c8fe5652ea5d95e5ce150bedde89624108ba4a 100644
--- a/mmaction/models/recognizers/recognizer2d.py
+++ b/mmaction/models/recognizers/recognizer2d.py
@@ -20,7 +20,8 @@ class Recognizer2D(BaseRecognizer):
         return loss
 
     def forward_test(self, imgs):
-        """Defines the computation performed at every call when testing."""
+        """Defines the computation performed at every call
+        when evaluation and testing."""
         batches = imgs.shape[0]
         imgs = imgs.reshape((-1, ) + imgs.shape[2:])
         num_segs = imgs.shape[0] // batches
diff --git a/mmaction/models/recognizers/recognizer3d.py b/mmaction/models/recognizers/recognizer3d.py
index 99e723322d1e77bc4d9fa5590ad14c56fc1bdd72..6179b892b440c350b791692fdd63e84073b2c1a2 100644
--- a/mmaction/models/recognizers/recognizer3d.py
+++ b/mmaction/models/recognizers/recognizer3d.py
@@ -18,7 +18,8 @@ class Recognizer3D(BaseRecognizer):
         return loss
 
     def forward_test(self, imgs):
-        """Defines the computation performed at every call when testing."""
+        """Defines the computation performed at every call
+        when evaluation and testing."""
         imgs = imgs.reshape((-1, ) + imgs.shape[2:])
 
         x = self.extract_feat(imgs)