Add initializer arguments to TransformerEncoder/TransformerDecoder

[mattdangerw]

We should add kernel_initializer and bias_initializer arguments to the TransformerEncoder/TransformerDecoder. These initializer should be passed to the dense and multi head attention layers with the encoder/decoder, and saved in the config.

[mattdangerw]

New contributors should feel free to pick this up! Ping if more details are required.

[aflah02]

Hey
I'd like to take this up. I'll have look at the relevant code base and get back if I face any issues

[mattdangerw]

Thanks!

[aflah02]

Hey
Just wished to confirm a few things

• I've used initializers from keras.initializers as that is what's used in all other places so I thought that's what we'll be using here as well
• I'm also adding some tests, is there anything else I should do too? I've added relevant docstrings to the encoder and decoder files

[chenmoneygithub]

I just dropped some comments on your PR, overall looks good. Thanks for your contributions!

[aflah02]

I think we can close this now @mattdangerw

[chenmoneygithub]

Thanks for your contribution!

[aflah02]

I had fun too :)