Let's put our newly gained knowledge to the test. Answer the following questions:

1. How does LSTM solve the vanishing gradient problem of RNN?
2. What are all the different gates and their functions in an LSTM cell?
3. What is the use of the cell state?
4. What is a GRU?
5. How do bidirectional RNNs work?
6. How do deep RNNs compute the hidden state?
7. What are encoders and decoders in the seq2seq architecture?
8. What is the use of the attention mechanism?