We've been referring to this package as "NengoLMU", which is kind of confusing because it doesn't really use Nengo at all. So we're changing the name to KerasLMU, which reflects the fact that this is a Keras-based implementation of LMUs. We're also renaming the package itself to align with this name (so people will now import keras_lmu and pip install keras-lmu). This is to clear the room for planned future implementations (e.g., a nengo-lmu which would actually be a Nengo-based LMU implementation). Although those packages don't exist yet, renaming the package itself at the same time as the "KerasLMU" name change seems appropriate.
Note that I'll be setting up a separate lmu package, so people will still be able to pip install lmu, and if they do that then import lmu will give them a deprecation message directing them to keras-lmu. In the future that lmu package will be used as a meta-package to install multiple LMU implementations.
We've been referring to this package as "NengoLMU", which is kind of confusing because it doesn't really use Nengo at all. So we're changing the name to KerasLMU, which reflects the fact that this is a Keras-based implementation of LMUs. We're also renaming the package itself to align with this name (so people will now
import keras_lmu
andpip install keras-lmu
). This is to clear the room for planned future implementations (e.g., anengo-lmu
which would actually be a Nengo-based LMU implementation). Although those packages don't exist yet, renaming the package itself at the same time as the "KerasLMU" name change seems appropriate.Note that I'll be setting up a separate
lmu
package, so people will still be able topip install lmu
, and if they do that thenimport lmu
will give them a deprecation message directing them tokeras-lmu
. In the future thatlmu
package will be used as a meta-package to install multiple LMU implementations.