add docs for tfutils.dependency

docs/modules/tfutils.rst

@@ -103,6 +103,14 @@ tensorpack.tfutils.export module
     :undoc-members:
     :show-inheritance:
 
+tensorpack.tfutils.dependency module
+------------------------------------
+
+.. automodule:: tensorpack.tfutils.dependency
+    :members:
+    :undoc-members:
+    :show-inheritance:
+
 Other functions in tensorpack.tfutils module
 ---------------------------------------------
 

docs/tutorial/faq.md

@@ -19,12 +19,13 @@ Then it is a good time to open an issue.
 ## How to print/dump intermediate results during training
 
 1. Learn `tf.Print`. Most of the times, adding one line in between:
-```python
-tensor = obtain_a_tensor()
-tensor = tf.Print(tensor, [tf.shape(tensor), tensor], tensor.name, summarize=100)
-use_the_tensor(tensor)
-```
-is sufficient.
+
+   ```python
+   tensor = obtain_a_tensor()
+   tensor = tf.Print(tensor, [tf.shape(tensor), tensor], tensor.name, summarize=100)
+   use_the_tensor(tensor)
+   ```
+   is sufficient.
 
 2. Know [DumpTensors](../modules/callbacks.html#tensorpack.callbacks.DumpTensors),
 	[ProcessTensors](../modules/callbacks.html#tensorpack.callbacks.ProcessTensors) callbacks.

tensorpack/tfutils/dependency.py

@@ -25,7 +25,7 @@ def dependency_of_targets(targets, op):
         op (tf.Operation or tf.Tensor):
 
     Returns:
-        bool
+        bool: True if any one of `targets` depend on `op`.
     """
     # TODO tensorarray? sparsetensor?
     if isinstance(op, tf.Tensor):
@@ -48,7 +48,7 @@ def dependency_of_fetches(fetches, op):
         op (tf.Operation or tf.Tensor):
 
     Returns:
-        bool
+        bool: True if any of `fetches` depend on `op`.
     """
     try:
         from tensorflow.python.client.session import _FetchHandler as FetchHandler