why the output of LSTM is zero

[nelson]

Dear all

I build  a program based on dynet and find the value of output vector is zero after calling forward function. I copy the code of building graph of my program the ./example/rnnlm/train_rnnlm.cc and everything is fine. so I guess the issue is caused by initialization??  but i strictly follow the process of initialization.

Could someone give me some suggestions?

Thanks a lot!!

[nelson]

Thanks for interruption! The problem has been resolved and it was caused by a manual initialization of all parameter.