When lstm run more than 2000 batches in first iteration, the error (float point exception) occurred! some print info as following:

I0909 17:38:59.281976 23954 TrainerInternal.cpp:162] Batch=2000 samples=128000 AvgCost=0.239344 CurrentCost=0.201116 Eval: classification_error_evaluator=0.0999063 CurrentEval: classification_error_evaluator=0.0798125
I0909 17:40:21.527977 23954 Tester.cpp:111] Test samples=25000 cost=0.184603 Eval: classification_error_evaluator=0.07068
.............................................................................................................................. .........................................................................................................................................................................................................................................................................................................................................................................................................................Current Layer forward/backward stack is
LayerName: lstmemory_0
LayerName: fc_layer_0
LayerName: embedding_0
LayerName: word
*** Aborted at 1473414360 (unix time) try "date -d @1473414360" if you are using GNU date ***
Current Layer forward/backward stack is
PC: @ 0x7f0182f0ab25 __ieee754_exp
Current Layer forward/backward stack is
*** SIGFPE (@0x7f0182f0ab25) received by PID 23954 (TID 0x7f017130c700) from PID 18446744071611394853; stack trace: ***
Current Layer forward/backward stack is
@ 0x7f018414d710 (unknown)
Current Layer forward/backward stack is
@ 0x7f0182f0ab25 __ieee754_exp
Current Layer forward/backward stack is
@ 0x7f0182f20b52 __GI___exp
Current Layer forward/backward stack is
@ 0x800235 hppl::tanh()
Current Layer forward/backward stack is
@ 0x587383 paddle::LstmCompute::forwardOneSequence<>()
Current Layer forward/backward stack is
@ 0x58788a paddle::LstmCompute::forwardBatch<>()
Current Layer forward/backward stack is
@ 0x581bdc paddle::LstmLayer::forwardBatch()
Current Layer forward/backward stack is
@ 0x58521a paddle::LstmLayer::forward()
Current Layer forward/backward stack is
@ 0x614b94 paddle::NeuralNetwork::forward()
Current Layer forward/backward stack is
@ 0x61efe6 paddle::TrainerThread::forward()
Current Layer forward/backward stack is
@ 0x621194 paddle::TrainerThread::computeThread()
Current Layer forward/backward stack is
@ 0x7f01832553d2 execute_native_thread_routine
Current Layer forward/backward stack is
@ 0x7f01841459d1 start_thread
Current Layer forward/backward stack is
@ 0x7f0182a3a8fd clone
/data11/dis_ml/deeplearning/paddle/bin/paddle: line 46: 23954 Floating point exception${DEBUGGER} $MYDIR/../opt/paddle/bin/paddle_trainer ${@:2}