Hello guys,
i want to create learn how to create simple LSTM. Do you guys have a guess where can i find some tutorial? Model_zoo is not sadly not enough for me, i need more commentary on that…
Hello guys,
i want to create learn how to create simple LSTM. Do you guys have a guess where can i find some tutorial? Model_zoo is not sadly not enough for me, i need more commentary on that…
Hey, check out : https://github.com/FluxML/Flux.jl/pull/1428#issuecomment-745801257
Thats not understandable for me…
One route to understanding is to take an end-to-end working basic model in Flux which uses the Chain()
, train!()
and data_loader()
functions, possibly a simple regression type working with 1-d data, reduce the model to a single Dense()
function (fully connected) and practice making minor tweaks to the code and explaining the result. When confident, introduce an LSTM()
function into the chain before the final fully connected layer and once that is working continue building the model to more complexity.
It’s more meaningful than copy and paste.
Presumably, it means that you can read Recurrent Models documentation which is not in the stable version of docs yet.
Here’s an example that I found helpful: https://github.com/mkschleg/FluxBooks.jl/blob/main/RNNs.ipynb