Created by: youan1

在paddle的配置文件里添加了计算两个slot的

emba_b_cos_sim=paddle.layer.cos_sim(name="emba_b_cos_sim",
                                    a=emba_pooling,
                                    b=embb_pooling,
                                    scale=1)
concat_layer =  paddle.layer.concat( name="concat_layer", input=[dense_data_input,, emba_b_cos_sim])

之前没有添加emba_b_cos_sim的时候

concat_layer =  paddle.layer.concat( name="concat_layer", input=[dense_data_input,])

dropout_layer0 = paddle.layer.dropout(input=concat_layer,  dropout_rate=0.2)

concat_layer2 = paddle.layer.fc(name="concat_layer2", input=[dropout_layer0], size=(dense_data_input_size + 1))

然后concat_layer2后面就开始接隐藏层了，唯一的变化就是在concat_layer里输入添加了一个emba_b_cos_sim，然后在concat_layer2里将size添加了一个 +1

我默认emba_b_cos_sim的size是为1的

可以正常运行

然后将emba_b_cos_sim输入到一个concat层，并作为上面的隐藏层的输入 本地调试时出如下错误： 请paddle的同学看看是什么问题

1016 17:01:59.181259  5025 Util.cpp:166] commandline:  --use_gpu=False
I1016 17:07:13.598457  5025 GradientMachine.cpp:85] Initing parameters..
I1016 17:07:59.232594  5025 GradientMachine.cpp:92] Init parameters done.
F1016 17:08:19.681207 18412 Vector.h:132] Check failed: start + size <= src.size_ (1179396 vs. 1172889)
*** Check failure stack trace: ***
    @     0x7f80d010ae6d  google::LogMessage::Fail()
    @     0x7f80d010e91c  google::LogMessage::SendToLog()
    @     0x7f80d010a993  google::LogMessage::Flush()
    @     0x7f80d010fe2e  google::LogMessageFatal::~LogMessageFatal()
    @     0x7f80d094f858  paddle::SgdThreadUpdater::threadUpdateDense()
    @     0x7f80d095053f  _ZNSt17_Function_handlerIFvimEZN6paddle16SgdThreadUpdater11finishBatchEfEUlimE_E9_M_invokeERKSt9_Any_dataim
    @     0x7f80d0081a0c  _ZNSt6thread5_ImplISt12_Bind_simpleIFZN6paddle14SyncThreadPool5startEvEUliE_mEEE6_M_runEv
    @     0x7f80cf8df8a0  execute_native_thread_routine
    @     0x7f80d9b801c3  start_thread
    @     0x7f80d91a812d  __clone
    @              (nil)  (unknown)
Aborted (core dumped)