Created by: hohaxu

按照在线文档里面的教程，试着训练了下paddle/demo/recommendation。不知道为什么我training loss比较教程里看起来要高很多? 用的paddle版本是4月30号的896b9c55, 没有做任何改动。

epoch 1 训练： cost=6.11816 在线教程： cost=3.32155 epoch 2 训练：cost=6.11527 。。。

＝＝＝ training log ＝＝＝ [INFO 2017-05-02 11:20:06,711 networks.py:1472] The input order is [movie_id, title, genres, user_id, gender, age, occupation, rating] [INFO 2017-05-02 11:20:06,711 networks.py:1478] The output order is [mse_cost_0] I0502 11:20:06.793532 18703 Trainer.cpp:165] trainer mode: Normal I0502 11:20:06.953569 18703 PyDataProvider2.cpp:243] loading dataprovider dataprovider::process I0502 11:20:06.992805 18703 PyDataProvider2.cpp:243] loading dataprovider dataprovider::process W0502 11:20:06.993029 18703 Trainer.cpp:622] --test_all_data_in_one_period was deprecated, since we will always do test on all test set I0502 11:20:06.993054 18703 GradientMachine.cpp:86] Initing parameters.. I0502 11:20:07.167628 18703 GradientMachine.cpp:93] Init parameters done. ................................................................................................... I0502 11:22:33.724074 18703 TrainerInternal.cpp:165] Batch=100 samples=160000 AvgCost=6.33338 CurrentCost=6.33338 Eval: CurrentEval: ................................................................................................... I0502 11:24:40.304360 18703 TrainerInternal.cpp:165] Batch=200 samples=320000 AvgCost=6.27258 CurrentCost=6.21178 Eval: CurrentEval: ................................................................................................... I0502 11:26:47.475955 18703 TrainerInternal.cpp:165] Batch=300 samples=480000 AvgCost=6.2448 CurrentCost=6.18925 Eval: CurrentEval: ................................................................................................... I0502 11:28:52.041764 18703 TrainerInternal.cpp:165] Batch=400 samples=640000 AvgCost=6.22759 CurrentCost=6.17594 Eval: CurrentEval: ................................................................................................... I0502 11:30:53.115355 18703 TrainerInternal.cpp:165] Batch=500 samples=800000 AvgCost=6.21779 CurrentCost=6.17862 Eval: CurrentEval: .................................................................I0502 11:32:13.507774 18703 TrainerInternal.cpp:181] Pass=0 Batch=565 samples=902826 AvgCost=6.21333 Eval: I0502 11:32:44.170591 18703 Tester.cpp:115] Test samples=97383 cost=6.11816 Eval: I0502 11:32:44.170732 18703 GradientMachine.cpp:64] Saving parameters to ./output/pass-00000 ................................................................................................... I0502 11:34:50.525712 18703 TrainerInternal.cpp:165] Batch=100 samples=160000 AvgCost=6.15603 CurrentCost=6.15603 Eval: CurrentEval: