* remove pick dependency

* fix bug

* small fix

* modify executor.py for save and load

* clean code

* Add usage example

* refine executor run function

* fix bug

* refine executor code

* fix block bug

* fix comments

* fix bug

* fix pass num

* aysnc send/recv, seriliaze/deserialize using threadpool

* implement paralell deserialization correctly

The English of the previous API is bad.

neural machine translation model support beam search with while op

* limit variable type to lod tensor in memory optimization transpiler

* refine policy

* support while operator

* fix random seed and training data order

* refine get_cfgs method to support multi while operators

* refine codes

* refine conv2d_transpose layer doc

* fix conv2d_transpose doc

* fix doc

* Updating the cluster trainign doc

* Fixed comments

* Updating few comments in recv_op

2. Fix evaluator using 'reduce_sum' op instead of 'mean' op

* also div num_samples when return cost of nce_op