Hi Ben, Thanks for writing. I have two concerns. You say "replace OCC with Hyperon". There's an ocean of stable, debugged, performance tuned, working code. When you call it "opencog classic", it suggests that you will be providing some kind of portability path for existing applications. Yet, I suspect that there aren't any compatibility plans.
The other concern is that by calling it "classic", you are sucking all the air out of development. There's only a finite amount of development talent. I would be much happier if you were recruiting developers from out of the ranks of Neo4J or tinkerpop or grakn.ai or some such projects, instead of cannibalizing opencog. Enlarge the pool; don't turn it into a zero-sum competition for developer mindshare. -- linas On Mon, Aug 29, 2022 at 6:25 PM Ben Goertzel <[email protected]> wrote: > > I'm motivated to provide (3) but sometimes dispense 1 & 2. I keep saying > "me", because Ben has pulled almost everyone from off of the projects here, > and onto other projects. > > A number of SingularityNET folks who were previously working on what > I've been calling "OpenCog Classic" (the version of OC Linas is > actively maintaining and developing) are now working on Hyperon, yeah, > see > > https://github.com/trueagi-io/ > > https://wiki.opencog.org/w/Hyperon. (not wholly up to date) > > or see the talks from AGI-22 Hyperon workshop at > > https://www.youtube.com/watch?v=BYvOMXl8zcc > > Linas, if you don't like "OpenCog Classic" as a label to use for "the > version Linas is actively maintaining/developing" it would be great if > you'd suggest another name... > > Hyperon shares largely the same conceptual foundation and high-level > design as OpenCog and on my end the AI algos I want to run on Hyperon > overlap very closely w./ the ones I wanted to run on OCC ... so I do > think it's valid to call Hyperon "a species of OpenCog" as opposed to > a totally new architecture > > In the big picture it's probably good to have multiple related > approaches/versions out there, exploring different regions of design > space... this has worked out OK e.g. in the operating system domain > obviously... > > As you'll see if you watch the AGI-22 talks (e.g. the ones by Alexey, > Nil and Jonathan Warrell), the MeTTa (fka "Atomese 2") interpreter (a > key part of Hyperon) now works well enough that hardy souls with a > taste for weird functional languages can play with it and write > interesting code... > > However, it is not yet mature enough to be recommended for practical > applications... documentation is incomplete, performance is not yet > optimized, and language features are still being tweaked/added based > on experimental usage > > We anticipate that once the framework matures, most development on > Hyperon will be done via AI-algo- or application-specific DSLs written > in MeTTA ... one thing we are now working on is a DSL (written in > MeTTa) for writing MeTTA DSLs. We anticipate MeTTA DSLs then being > invoked within code in other languages (python, Haskell, Julia) much > as ML/DL frameworks are now invoked w/in scripts... (See Adam > Vandervorst's talk at AGI-22 for a high level overview on this...) > > Chatbot-wise, the trajectory we're on is to > > -- open-source the dialogue-system framework we've written for the > Grace eldercare robot (awakening.health), which uses OpenCog Classic > for some things along with a bunch of transformer NN models > > -- replace OCC with Hyperon in the above framework ... and also > introduce an experimental usage of Hyperon for episodic memory > associated w/ dialogue > > However, the above has not been done yet meaning there is no > Hyperon-related "chatbot" framework you can use at this. moment.. > > The 3 main use-cases we're initially looking at for Hyperon, to pursue > before the system is mature as guides for development, are > agent-control in Minecraft (portable in many ways to other virtual > worlds), dialogue-systems as noted above, and genomics (the OCC > BioAtomspace should be portable to Hyperon fairly straightforwardly) > > -- Ben > > -- > You received this message because you are subscribed to the Google Groups > "opencog" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/opencog/CACYTDBdtM8B02XrjH2c6k5x40GhDh1%3DZfEuZ_pK%3DcbUkSPV0zQ%40mail.gmail.com > . > -- Patrick: Are they laughing at us? Sponge Bob: No, Patrick, they are laughing next to us. -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAHrUA36L%3D3JYiUXOk2SME_j7PXGfEUrqc00TPQqc2wjtuvi1qQ%40mail.gmail.com.
