* Change the transposed conv2d initializer.

* small fix.

* Add more model configure, such as AlexNet, VGG, GoogleNet, DPN, ResNet etc.
* Add download_imagenet2012.sh
* Update doc.
* Add Chinese doc.

* DQN Atari based on RLLab

* DQN Atari based on RLLab

* DQN Atari based on RLLab

* refactor code without RLLab framework

* add module of saving and loading policy model

* refactor code structure, add DoubleDQN and DuelingDQN modules

* add fluid argmax, flatten utils

* update README.md

* udpdate replay memory

* udpdate replay memory

* update readme, code clean

* clean code and fix codestyle

* fix codestyle

* update README.md

* revisions|history->context, randint->random

* revisions| add comment for max-pooling operation in atari

* update wiki

* refine wiki

* train.py default args

* update wiki cn

* update wiki

* Add head bbox for Pyramid-Box model.

* Fix bug in transform_labels and satisfy_sample_constraint function.

* Add object_coverage constraint & fix bbox_labels data layout bug

* code clean

* Support groups in de-conv and fix bug.

* Clean code.

* Fix model config.

* Some mirror changes in data argumentations.

* Set learning rate by input arguments.

* Add check for model_dir and gitignores

* code clean

* code clean

* Support groups in de-conv and fix bug.

* Clean code.

* Fix model config.

* Support groups in de-conv and fix bug.

* Clean code.

* caffe2fluid: fix bug in transforming Scale layer without bias parameter

* caffe2fluid: fix bug about Scale layer transformation and model saving

* training code of pyramidbox

* Fix code format.

* load pretrained vgg model

* load pretrained vgg model

* delete irrelevant tools

* delete unused lines

* PyramidBox for face detection.

* Update some hyper parameters.

* Update pyramidbox.py

support multiple data layer;
fix bug about caffe_pb2;

* [DQN]source code commit

* Update README.md

* Update README.md

* add mountain-car curve

* Update README.md

* Update README.md

* clean code

* fix code style

* [fix code style]/2

* remove some tensorflow package

* a better way to sample from replay memory

* code style