Re: Modifying the TD Code in PDP++

["Randall C. O'Reilly" <oreilly@grey.colorado.edu>]

Ok, I was inspired to make my current TD code work outside the BG/PFC
framework and tested it out on the simple TD example from my
"Computational Explorations.." textbook, chapter 6.  Everything seems
to work fine, and as an extra bonus I added the 3 lines (in the right
places.. ;) to implement the Daw & Touretzky
subtract-the-average-reward variant.  Just click on the sub_avg
button in the rew field in the RewPredLayerSpec...

the updated code is in an updated version of this file:

ftp://grey.colorado.edu/pub/oreilly/pdp++/pdp++_2.3a_src.tar.gz

and I put the TD example in:

ftp://grey.colorado.edu/pub/oreilly/misc/lba_rl.tar.gz

just run new_rl_cond.proj.gz with the new version of leabra++ from the
src tar above.

Cheers,

				- Randy