A modern programming language that makes developers happier.

kotlinlang

Is there somebody already working on attention models or/and LSTMs ?

I tried to implement LSTM units on TF Java API 1.15 from scratch via basic operands but found a few operands like LSTMCelll in TF Java API 2.x, decided to wait before the update on 2.x. So, attention is a new thing in frameworks and implemented mostly in Python, have no tries yet to implement it in Kotlin

I am just learning about attention models and the best way to learn it is to implement it. AFAIK on the end its just matrices multiplications, but to have a usefull use case, they need encoder and decoders with LSTMs. Feel free to cerrect me if I am wrong

Not necessarily: <http://nlp.seas.harvard.edu/2018/04/03/attention.html|http://nlp.seas.harvard.edu/2018/04/03/attention.html>

When I have some time I will reproduce the above in Kotlin with kmath-torch for you ta have an example

It will be interesting to see your implementation