Any chance of GRU/LSTM layers ?

crystal-data / num.cr

Scientific computing in pure Crystal

MIT License

151 stars 12 forks source link

Any chance of GRU/LSTM layers ? #43

Open clive-g-brown opened 4 years ago

clive-g-brown commented 4 years ago

Code Sample, a copy-pastable example if possible

# Your code here

Problem description

[this should explain why the current behaviour is a problem and why the expected output is a better solution.]

christopherzimmerman commented 4 years ago

I'd love to add these in. Primary issue is figuring out how to back propagate along layers that need more than one gradient. This will probably require more Num::Grad work than it will Num::NN work.