nengo / keras-lmu

Keras implementation of Legendre Memory Units
https://www.nengo.ai/keras-lmu/
Other
207 stars 35 forks source link

Add new computational path when return_sequences=False #53

Open NarsimhaChilkuri opened 1 year ago

NarsimhaChilkuri commented 1 year ago

When return_sequences=False and the hidden cell is not an RNN, the LMUFeedforward computation can be made more efficient by computing the output as described in equation (25) in https://arxiv.org/pdf/2102.11417.pdf.

TODO: