Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Nov 13 2018 07:41
    AlyeskaBear commented #48
  • Nov 12 2018 21:38
    brohrer commented #48
  • Nov 12 2018 06:49
    AlyeskaBear closed #48
  • Nov 12 2018 06:49
    AlyeskaBear edited #48
  • Nov 12 2018 06:09
    AlyeskaBear edited #48
  • Nov 12 2018 06:08
    AlyeskaBear opened #48
  • Oct 01 2018 18:03

    markroxor on master

    travis badge (compare)

  • Oct 01 2018 14:43
    brohrer commented #45
  • Sep 26 2018 13:00
    markroxor commented #45
  • Sep 26 2018 12:54

    markroxor on travis

    (compare)

  • Sep 26 2018 12:53

    markroxor on master

    travis, tox and flake8 fix ignore tox files and logs (compare)

  • Sep 26 2018 12:51

    markroxor on travis

    travis, tox and flake8 fix ignore tox files and logs (compare)

  • Sep 26 2018 12:37
    markroxor commented #47
  • Sep 26 2018 12:11
    markroxor added as member
  • Sep 26 2018 12:10
    markroxor commented #45
  • Sep 26 2018 11:43
    brohrer commented #45
  • Sep 26 2018 10:52
    brohrer commented #47
  • Sep 26 2018 10:44

    brohrer on master

    adding logger Merge pull request #47 from mar… (compare)

  • Sep 26 2018 10:44
    brohrer closed #47
  • Sep 26 2018 10:44
    brohrer closed #29
AYF
@realazthat
@brohrer oh oh does becca only support python 2.X ?
automenta
@automenta
https://github.com/automenta/beccapy3 it nearly ports to python 3. just a little more work was required
there is a python3 script which translates python2 to python3 thats what i tried her
here
of course if you settle for python2 what is the point of python3. thats what i cant figure out
AYF
@realazthat
they are two languages
the point of python 3 is to upgrade certain non-backwards-compatible features
and python2 is to support the large number of legacy libs
thanks for the link
automenta
@automenta
its still python. it still warns me about the tab and spaces stuff
er, error not warning
i guess they couldnt manage to get a parser upgrade in v3
Brandon Rohrer
@brohrer
@realazthat very cool that you are using becca! Yes, I've done some work under becca 4 with audio and video. The biggest trick is creating the sensors. For audio i tried a bunch of things and settled on short-windowed ffts, then binned those into a dozen or two bins, then filtered the levels of those bins to smooth them in time. All the code for it is in the becca_world_listen github repository.
I used ffmpeg to break apart the videos into stills and to reassemble stills into videos for all of my YouTube demos. For the dashboards, i created each frame from scratch using matplotlib then stitch them together using fmpeg. The code for that is all in the becca_control_panel repo.
Brandon Rohrer
@brohrer
Becca is all 2.x right now. I have tried write all my new code to be 3.x compatible, like new style print statements, but i haven't done the work to officially port it. I expect there may be some issues with the Anaconda packages I'm using, but i haven't dug through it yet.
@automenta is currently the expert on 3.x compatibility 🙂
Brandon Rohrer
@brohrer
Here are the links:
I haven't updated the control panel, watch or listen repositories since becca 5, so I don't think they'll actually run. The code will show what I did, though, if you have any detailed questions about it.
AYF
@realazthat
thanks
i'll prolly look at things more indepth late next week
Brandon Rohrer
@brohrer
I just hit a roadblock in the development of Becca 8. The approach I'm using estimates reward for state-action pairs (similar to a traditional value function). I realized I need to revert back to the approach in Becca 7 where I used state-action-state tuples as my basis. This allows for state predictions, world modeling and flexible multi-step planning. Becca 8 taught me several things about how to best represent and combine inputs. Now I need to move on to Becca 9 and integrate the best of both 7 and 8. I'll release the current state as Becca 8.1 and move on.
Brandon Rohrer
@brohrer
Very cool @automenta. It's fun to watch OpenNARS in action.
automenta
@automenta
thank you yes even if it's just random actions its still nearly machinima
Tapo
@btapo
@automenta the implementations are really cool
@brohrer congratulations on getting out another BECCA version
AYF
@realazthat
very nice
AYF
@realazthat
things are happening :)
Brandon Rohrer
@brohrer
Yes they are! Becca is ripe for a new release. I've re-engineered the reinforcement learner and it's working pretty well. Much better than any of the previous versions :-)
AYF
@realazthat
awesome
Brandon Rohrer
@brohrer
It's official. Becca 9 is out. It's performing better and running faster than ever. I'm having a celebratory Scotch.
AYF
@realazthat
haha
:thumbsup:
:beer:
Tapo
@btapo
Great. Cheers
automenta
@automenta
=D
Brandon Rohrer
@brohrer
Thanks all!
Good idea @realazthat. A beer, too.
automenta
@automenta
https://github.com/google/grumpy Grumpy is a Python to Go source code transcompiler and runtime that is intended to be a near drop in replacement for CPython 2.7. The key difference is that it compiles Python source code to Go source code which is then compiled to native code, rather than to bytecode. This means that Grumpy has no VM. The compiled Go source code is a series of calls to the Grumpy runtime, a Go library serving a similar purpose to the Python C API (although the C API is not directly supported).
borasy
@borasy
Hi I'm new here. I saw the video about Becca detecting movie trailers while watching The Big Bang Theory, and I wanted to use Becca to do a similar thing. How should I start? Thanks
Brandon Rohrer
@brohrer

Hi @borasy. Welcome! Unfortunately the because code is in a very researchy state right now, and also unfortunately I don't currently have the bandwidth to provide an introduction. Sorry I can't provide more help at the moment!

Brandon

borasy
@borasy
Oh, I understand. That's alright. I'll try to test out the codes myself first then.
automenta
@automenta
https://www.youtube.com/watch?v=J9atPnbswMQ RL test environment. the video has random dummy agents. clients connect through UDP https://github.com/automenta/failchamber