name correction in validatestatprinter

README.md

@@ -1,12 +1,11 @@
 # tensorpack
-Neural Network Toolbox based on TensorFlow
+Neural Network Toolbox on TensorFlow
 
-In development. No document, don't use.
+In development. No document.
 
 ## Features:
 + Scoped abstraction of common models.
-+ Provide callbacks to control training behavior (as in [Keras](http://keras.io)).
++ Callbacks systems to control different aspects of training.
 + Use `Dataflow` to gain fine-grained control on data preprocessing.
-+ Automatically use the Queue operator in tensorflow to speed up input.
 + Training and testing graph are modeled together. Just need to follow the conventions to setup stuffs.
-+ Use tensorboard easily.
++ Write summary easier for tensorboard.

tensorpack/callbacks/validation_callback.py

@@ -9,6 +9,7 @@
 
 from ..utils import *
 from ..utils.stat import *
+from ..tfutils import *
 from ..tfutils.summary import *
 from .base import PeriodicCallback, Callback, TestCallbackType
 
@@ -82,7 +83,8 @@ def __init__(self, ds, names_to_print, prefix='validation', period=1):
         self.names = names_to_print
 
     def _find_output_vars(self):
-        self.vars_to_print = [self.get_tensor(n) for n in self.names]
+        self.vars_to_print = [self.get_tensor(
+            get_op_var_name(n)[1]) for n in self.names]
 
     def _get_output_vars(self):
         return self.vars_to_print