Training accuracy can't be augmented

[pham_su]

Can someone point it out for me why the accuracy for my training model is so low? As you can see, I tried using regularizer, Dropout, simple model with one bidirectional LSTM, I also tried 100, 200 units for LSTM and I tried changing larger learning rate. It is still not working.
And, I tested all of my functions, it passed all the tests.
Many thanks!
Here is my model

[code removed - moderator]

[balaji.ambresh]

The learning rate is high. I recommend you start with the default learning rate for adam.

[pham_su]

Many many thanks. It worked now!