update readme

README.md

@@ -3,12 +3,15 @@ Neural Network Toolbox on TensorFlow
 
 Still in development, but usable.
 
-See some interesting [training examples](examples) to learn about the framework:
+See some [examples](examples) to learn about the framework.
+You can actually train them and reproduce the performance... not just see how to write code.
 
 + [DoReFa-Net: training binary / low bitwidth CNN](examples/DoReFa-Net)
-+ [Double-DQN and A3C for playing Atari games](examples/Atari2600)
-+ [ResNet for Cifar10 classification](examples/ResNet)
 + [IncpetionV3 on ImageNet](examples/Inception/inceptionv3.py)
++ [ResNet for Cifar10 classification](examples/ResNet)
++ [Fully-convolutional Network for Holistically-Nested Edge Detection](examples/HED)
++ [Double-DQN plays Atari games](examples/Atari2600)
++ [Batch-A3C plays Atari games with demos on OpenAI Gym](examples/OpenAIGym)
 + [char-rnn language model](examples/char-rnn)
 
 ## Features:

examples/DoReFa-Net/README.md

@@ -30,7 +30,7 @@ export PYTHONPATH=$PYTHONPATH:`readlink -f tensorpack`
 
 ## Support
 
-Please use [github issues](https://github.com/ppwwyyxx/tensorpack/issues) for any issues related to the code.
+Please use [github issues](https://github.com/ppwwyyxx/tensorpack/issues) for any issues related to the code itself.
 Send email to the authors for general questions related to the paper.
 
 ## Citation

examples/ResNet/cifar10-resnet.py

@@ -21,9 +21,9 @@ This implementation uses the variants proposed in:
 Identity Mappings in Deep Residual Networks, arxiv:1603.05027
 
 I can reproduce the results on 2 TitanX for
-n=5, about 7.1% val error after 67k step (8.6 step/s)
-n=18, about 5.7% val error (2.45 step/s)
-n=30: a 182-layer network, about 5.6% val error after 51k step (1.55 step/s)
+n=5, about 7.1% val error after 67k steps (8.6 step/s)
+n=18, about 5.8% val error after 80k steps (2.6 step/s)
+n=30: a 182-layer network, about 5.6% val error after 51k steps (1.55 step/s)
 This model uses the whole training set instead of a train-val split.
 """
 

tensorpack/train/multigpu.py

@@ -9,6 +9,7 @@ from six.moves import zip, range
 
 from ..models import TowerContext
 from ..utils import logger
+from ..utils.naming import *
 from ..utils.concurrency import LoopThread
 from ..tfutils.summary import summary_moving_average
 from ..tfutils.modelutils import describe_model