Go implementation of Leabra algorithm for biologically-based models of cognition, based on emergent framework (with Python interface)
No new features -- just updated GUI etc infrastructure.
Lots of new updates here, including Python versions of most of the models too.
Major updates, this is release being used for CompCogNeuro/sims release, including python and go builds.
UnitVarIdx
and SynVarIdx
get index for variableUnitVal1D
and SynVal1D
is fast index access used by all routines -- these are only ones that need to be updated in subtypes.Here's what the new code looks like in MemStats for examples/hip/hip.go
, line 706
:
actMi, _ := ecout.UnitVarIdx("ActM")
targi, _ := ecout.UnitVarIdx("Targ")
actQ1i, _ := ecout.UnitVarIdx("ActQ1")
for ni := 0; ni < nn; ni++ {
actm := ecout.UnitVal1D(actMi, ni)
trg := ecout.UnitVal1D(targi, ni) // full pattern target
inact := ecin.UnitVal1D(actQ1i, ni)
ra25 now runs (mostly) clean on race detector with lots of pushing of random buttons..
GoGi GUI framework now stable.
Updated python makefile and ra25 for latest gopy update.
all code that includes mat32 needs to change goki/gi/mat32 -> goki/mat32
Everything is now sufficiently feature-complete and well-tested to warrant a 1.0.0 release.
Everything is now far enough along that the main demo project in leabra/examples/ra25 can be used as a starting point for creating your own projects!