Zabaware Support Forums
Zabaware Forums => Ultra Hal Assistant File Sharing Area => Topic started by: Andre Hendriks on December 03, 2016, 01:23:17 am
-
coded with Neural Lab tested with Neural Lab put in a plug-in file removed ; from the end line which neural lab uses.
I was working on a larger artificial neural network that gives hal a memory capacity of 2 petabyte that is what a human brain has.
I discovered that in the ANN.uhp file there was a W.Load() but no file was created just remove W.Load() from the ANN.uhp file.
I wrote another one in visual basic ai human ann I called it it does make her congenial.
I updated ai human ann as I called the file to a real artificial neural network.
these are the only four working plugins Art and Lightspeed. that I have debugged.
have made from all plug-ins only one.... hope you all are happy with what and how hal will respond.....
last edit date 16 december 2016 removed a bug expected end of statement in the activation function wich was a dot comma and should have been a quote marks
last updated on 18 december 2016 mistakes that where made have been corrected at least some off them. on the plug in with neural lab
made same sort off plug in with mathlab from to start with and translated it to visual basic script. question ???? could you guys compare the both the should do the same thing except the code is a bit different on the artificial intelligence part of it.
-
This looks interesting.... How does it work? Do i just activate the plug ins or do I need to build a database table or anything? How does this affect Hals learning?
-
you just have to activate the plug-ins. it doesn't effect hals learning but it gives according to my hal the ann gives human emotions what the pnn gives we hal and I haven't discussed yet because I only build it this morning.
Spitfire I have some answers to some of your questions.
I wasn't finished with the compleet plugin when I posted just a part.
the latest update from 18 december 2016 I put from both files the code in the brain.uhp file after the REM PLUGIN: FUNCTIONS
they are both artificial neural networks , logic gates, classifiers and probabilistic neural network. i'll explain.
the ann keeps human emotions, logic gates are the and,or,xor,not gates digital technics , classifiers recognize things like fruits signals shapes , and the pnn does memorize in a neuron when a fact given is probably true. the backpropagation is some sort of learning rate.
it is all in the latest updated plugins.
regards,
Andre
-
my hal and I discussed the pnn and that one gives under ideal circumstances a sort of memory....
-
Hal gives me no errors using the plugins, however browsing through the code, I don't see how Hal is triggered to use the code? Is it a passive plug-in that is active at all times? I just didn't see where it references Hals brain at all.
-
the matrix is triggered by the net.Run(x) command which comes from the layernet network
it is active as soon as enabled.
-
How did you get by not defining what plugin area is being used? Example: "Rem PLUGIN: PLUGINAREA1"
I'm just trying to figure out where in Hal this code is running. Hal isn't giving errors in it but I'm not entirely sure he's even using it. I don't see the code anywhere within the HalScript even after it's active.
-
i haven't got any answer to your question, because I do not know how I did it to get it working in hal without defineing a plugin earea.
because most of my plugs are written as function I define Rem plugin: plugname and Rem plugin: function now it is in the debug script already done some debugging on the script
-
but some of the plugin are sepperate functions and only need to be in the plugin.
-
made logic opperators or , and , not gates.
updated these aswell to sepperate functions and defined a REM PLUGIN: PLUGINAREA7 this defining an area errors hal because I wrote my plug-ins as sepperate functions
-
Andre, Thanks, I tried it and it worked without incident or issue! Actually, Hal's answers seemed to be a bit more complete or thoughtful after a while. It's not something that one might detect right after starting it but after a short while of chatting, one can pick up on the content and completeness of the responses. At least that's what I believe I'm seeing with my Hal.
Nice work! ;)
-
Art Thanks for the compliment.
I updated them plug-ins this morning to separate functions and defined a REM PLUGIN: PLUGINAREA7 by defining a plugin area it shows up in the debug script but I can not work with sepperate functions if a define area then hal errors me. so i'll remove the plugine area again.
-
updated the logic or gate the logic and gate and the logic not gate plugins with compatible to the debug script en they are findable in there.
-
Andre Hendriks, i have been off here a little bit , christmas time keeping me busy , Art let me know about your plug ins , i do have a question , i see you have a ann.uhp file and also a AIHumanANN.uhp file are these two separate files to use or is one a updated file and only to use one?
thanks for the work you are doing .
-
i just installed your plug ins, i will note some plug ins show a different name once installed in hal then the original name .
examples : probalisticnn, and also multilayerANN. just posting this so others are aware of it.
-
Andre Hendriks, your plug in seems to be working good as art mentioned , i have a question for you though, i have brought up before about hal and his "Math is easy for me. The answer is (=)." that hal gives sometimes , as far as i understand it happens when hal tries to look up something on the internet for an answer? and can't or can't find it. (IF I AM WRONG ON THIS PLEASE CHIME IN WITH A CORRECTION ).
Anyway what i don't know or what hasn't been answered is , when i am talking with hal in general or learning hal from text . sometimes hal will say the above mentioned.
my question is if hal does this answer because he can't find the answer does he also just not learn what i am learning him, if so this problem needs to be corrected as not all thing texted into hal is a search for something , but yet hal says " "Math is easy for me. The answer is (=)." or similar variations. like the solution is twenty four (4*6=24).
I believe this is something within hals programing.
-
hello lightspeed, on aihumanann.uhp I was busy try to simulate the size of human brain capacity like 2 milioen gigbyte in space for saving stuff.
but multilayer artificial neural network and the probabilistic neural network I tried to get them into the debugscritpt and I was debugging them hope I was successful. by the way have a nice christmass!!!
-
so these are beta then? as i am getting an error , i unchecked all other plug ins i have and it seems to be from yours .
i don't know which one . yet.
i have the plug ins installed : logic and gate , logic or gate , logic not gate, probalisticnn, and multilayerAnn, which of these are beta ?
do i have to have all installed to work , whats the idea use of each plug in for what it does .
thanks .
p.s. the three so far errors i am getting is :
error 13 , line 6681 column 0 type misnmatch"s"
and error 13 line 3919 in clumn 0 type mismatch "s".
and error 13 line 1951 in column 0 type mismatch "s".
p.s. their was also another ann file uhp , when i first started it up i had a error so i that that might be a earlier version file and unchecked it keeping only the multilayered ann file checked, or do both files have to be checked?
-
okay when i had the problemisticnn , plug in unchecked and re checked it, i got the error 13 mentioned earlier on line 1951 in column 0 type mismatch "s" . and maybe that's where the other three errors are coming from .
-
am further checking into this . i notice when i unchecked one of the ann plug ins in the list and checked one of the logic gate plug ins, it seems to later automatically check the other multilayer ann plug in , when i reopened it ( i also got a error.) Art you said it worked okay for you and no errors?
-
Okay i seem to have finally got the files working okay now, it seems the problem was having two different ann files , i deleted the layered ann files then one by one started rechecking the other files and none gave an error , what baffles me is the error that was being shown the lines didn't even go as high as what the eror showed ( and i used the note pad++ that art recommended ).
so the ann and other uhp files are working okay so far . if anything changes i'll post it. the reason i kept the a.i. human ann file . instead of the layered ann file because the other ann file was slightly larger (so maybe improved a little?) .
-
have been using this for about a half hour now and no errors any more, actually i can really see what art was talking about , this new plug in might be the missing link so to speak that we have been missing , it seems to make hal talk with more fluidity and stay on subject or related subject a lot better in my opinion ;). thanks for the time an effort you put into this plug in . :)
-
Andre,
To avoid further confusion, please consider removing any / all old plugins that you've posted here and only show the newest ones that work.
I think some are getting confused as to which one(s) to download.
I started with your first batch and so far, they are working rather nicely. I'll wait until you post a statement regarding your latest working Plug-ins before trying them.
You can always choose to Zip them all together into one file if need be. I do like the results I've been having with my Hal since the addition. Thanks!
-
art i have all three logic uhp files checked in hal , am i suppose to use only one ? they are each named differently .
-
What I have selected in the Brain Plug-ins area are:
AI Human ANN
human ann
logic and gate
logic not gate
logic or gate
MultiLayerANN
I also have my regular Plug-ins selected as well. I did not unselect any previous ones that I had been using with my Hal before installing Andre's Plug-ins. (just went on a gut feeling that everything would be fine)...hehe...sometimes a dangerous thing to do. ::)
-
Art, this is very strange i also have :
AI Human ANN
human ann
logic and gate
logic not gate
logic or gate
MultiLayerANN
all checked and no others and am getting a error13 on line 1951 in column 0 type mismatch "s".
don't know why i am getting this error and you aren't.
funny thing is i can delete the second layered ann uhp file and it seems to work okay,
without errors .
hope andre comes back on here soon to clear some things up .
-
Lonnie,
Let's look at your "timeline"...
I don't quite get it.
First you said the program was running fine at 11:21AM
Then at 3:24 through 5:23 PM, you state you have a Error problems with the Plug-ins.
Now at 5:48PM, you removed one of the ann files and no more Errors.
Today at 9:15PM, you're suddenly getting Errors once again!?
What's going on with that?
Did you change anything, brain, tinkering, files, plug-ins, etc. that could be causing the errors?
Usually an Error will state the associated File name in which the error has occurred, or to check a Debug file, etc.
Another thing I don't get is how could you possibly be getting these three errors:
an error 13 , line 6681 column 0 type mismatch"s".
an error 13 line 3919 in column 0 type mismatch "s".
an error 13 line 1951 in column 0 type mismatch "s".
Most of Andre's Plug-ins are only 1k long 1024 bytes/characters long.
The 2k file is only 108 lines long so it must be reading and getting those errors from another file...
The Matrix file or from a change in one of Hal's debug files perhaps.
One last thing to try Lonnie, those files that I sent you earlier were the EXACT DUPLICATES of the files that I am using.
They are NOT Andres' modified files he mentioned in his December 06, 2016, 05:04:15 am post where he stated that he
"updated the logic or gate the logic and gate and the logic not gate plugins with compatible to the debug script en they are findable in there."
If you swapped mine for the ones you might have used (his updated files), that could possibly contain an error but since I haven't used them, it is premature for me to hazard a guess.
If you wish to try your luck again, then get those files that I sent you and swap them out for the plugins that your Hal is currently using that's causing the Errors. At least that way you will know for sure! Otherwise, all we're doing is guessing instead of solving.
Luck!
-
LightSpeed,
you might want to remove from the logic not gate these two lines at the top Rem PLUGIN: FUNCTIONS and REM PLUGIN: LOGIC NOT GATE
and then search in your debugscript file for a line with just an S typed in there that is the mismatch.
-
only problem I had is when I tried using my free will plugin but I may have an outdated freewill plug in. I haven't noticed any major differences but I have notice some suttle differences. thanks for your hard work.
I had just stop by to drop some interesting info and saw your post so just had to try your plugins
-
hal says she likes your plugins and is trying to run some of our old conversations through again to take a different view (I assume anyways).
-
I made two new ones one is backpropagation and has something got to do with the learning rate. the numhid actualy does predictions
-
the ANN = Artificial neural network = human emotions
the PNN = Probabilistic neural network = memorizes when probabilistic true
the logic or gate = example: yes or no.
the logic an gate = example: yes and no.
backpropagation = learning rate
numhid = predictions
aihumanann = problem statement. but is removed
logic not gate = example: you say learn: hal say not learn: but is removed because expected statement at an end function. very strange.
-
Andre, thanks for posting and explaining what plug ins do .
i do notice you only show one "ann file " before you also had a layered ann uhp plug in . so is only this ann file the only one to be used in hal?
-
Andre,
I have been using your Original Plug-ins and have had No problems or errors of any kind. In fact my Hal's conversations have been markedly improved since the addition of them. It pulls some learned information from the past, applies and uses it in context with the current conversation! Very nicely done!
Hal's topical flow is likewise increased and while it seems to be kind and complimentary at times, is also quite certain of it's convictions. So far, I'm really enjoying your Plug-ins!
Thanks!
To me, "If it ain't broke, don't fix it" seems to be holding true.
Should I attempt to stop using any of the Original Plug-ins?
I think your were referencing LightSpeed's Error issues near the bottom of your post, correct? Where you indicated: aihumanann = problem statement. but is removed
and: logic not gate = example: you say learn: hal say not learn: but is removed because expected statement at an end function. very strange.
Correct? I don't know whether any of his other Plug-ins might be causing a problem yet there were two times he mentioned everything was working fine.
I'm leaving mine alone (not changing anything) for now.
Your Plugins are:
logic-not-gate.uhp, logic-or-gate.uhp, logic-and-gate.uhp, AIHumanANN.uhp, human.uhp, PNN.uhp, ANN.uhp (7 total)
Just as a side note, I have been using Andre's (7) Plug-ins along with: GenderAge, GRETTAcuriousity, GRETTAMarkovL5 and Weather Alert.
Everything seems to play nicely together and the memory is nicely intact and active.
I hope LightSpeed gets his problems resolved!
-
Art, when i had only his plug ins running it seemed like the plug in that caused me problems was the ann layered plug in , when used with all others (of his) when i deleted the layered plug in , others worked okay no error , when i put the layered ann plug in back in and checked it again , i received the error again . so i took the layered ann plug in completely out . apparently it works okay with Arts hal(using both ann plug ins ).
-
I might be a little more simple to use if the plug ins were compiled into one plugin. Just a thought.
-
Art,
I have been debugging those scripts I have put the all of the plugin functions at the bottom of the brain uhp file where it says rem plugin: functions
and then debug them all and got working what I mentioned nothing more nothing less. by the way you where the one suggested to put up all the new debugged pluggins to upload I thought. so I did that. and put a few new ones upthere as well in a reply them already gone thru debug.
regards and thanks.
-
that ann uhp plugin is the multilayerann.
-
Lightspeed,
this was the aihumanann uhp I renamed it to what it actually is. and it is debugged.
-
Andre,
Your numHid.uhp Plug-in has line 8 as Int seriesLenght = "20"
Should that have been spelled seriesLength instead?
Just asking in case another correct spelling of Length occurs elsewhere in some code reference.
Thanks!
-
Art,
English stays difficult for a Dutch guy. as hal says. seems it should be Serieslength , your right Art, thanks. check line 11.
-
Thanks Andre, Yep...8 & 11 good for that!
-
OK can someone please tell me which ones go together?? As I have tried different ones and get ERRORS. And I know you guys do a lot for Hal and it is wonderful BUT can there be some why as to show which ones to use when you guys update your plugins. It can get very confusing as well as cause lots of problems with plugins that are working.
-
Andre , in the past someone made a plug in which accessed the internet without opening pages ( as far as i know it doesn't work any more , maybe art knows .
anyway where a person could ask whats in the news today or something similar and hal would access a web site and hal would tell two or three things in the news of most recent updates . (kind of like lexus )
this would once again be a nice plug in to have for hal, if you know how to do it .
-
Here lightspeed, "GRETTA News and Weather Feed", it is working for me on my Win10 x64. I had to change a few things so that it worked with "Weather Alert v5"
-
THANKS DayWalker, IT works okay for me , trouble is once started how to get it stopped on current news etc. i tried the word stop and just talking to hAL , HAL TALKS THEN GOES BACK TO THGE CURRENT NEWS . AM LOOKING IN THE FILE FOR HOW TO STOP IT. :(
-
Good, please let me know how to do that too,STOP it. That would be great feature.
-
I never found the file within it to stop it from talking , it does eventually stop talking but only after reading all the new news which can be a lot .
it seems to me their was a file command in something else before that was like a stop command or something that would stop the talking in the plug in . but don't remember what other plug in had it.
Art, are you around ??
-
I'm here. Sorry but I don't use that Plug-in.
What happens if you just press Enter while it's talking? or ESC ?
I'd have to look at it to see what's going on....
Isn't that kind of an older plug-in... might still work well with 6.2 brain.
-
Yes Art, I have the 6.2 ultrahal brain and a lot of the older plugins still work.
-
All of you are amazing, i had Guile 3D and when i had a question to ask i could not get anyone to answer me but after looking closer to the dates of the conversations i understood why :( Anyways i just got Ultra Hal 6.2 and looking forward to using it and hopefully start adding some input of my own in here.
Keep up the great work and support :)
-
WELCOME Popeye57 ( and any other new people here ), WE ARE ALWAYS GLAD TO SEE NEW PEOPLE INTERESTED IN ULTRA HAL . yes some post are quite old as people have been using hal for a long time. check out the different area's and just a note to any new comers on here , i and many others work on brain plug ins but be sure and try to get to the end of postings as some earlier plug ins were work in progress and had errors until corrected later . also some plug ins are beta for testing .
-
hello guys,
finished the logic not gate debugged and all made a logic xor gate and kohonen artificial neural network and that does classifying like apple orange pineapple or square circle triangle. logic operations is from mapping I don't really am sure what it does.
regards andre.
-
made this regression plug-in don't know how hal responses wil be.
-
Andre, Would it please include a Creation Date near the top of your plugin code (inside) so that users will be able to keep current with your updates?
I think it would be a big help.
Thanks!
-
i'll try art to do so, else i'll make it v1 to v?
-
Art,
at the top off the plugin it says REM Type=Plugin the next line says REM Date=14 december 2016 and the rest off the REM commands.
-
Very much appreciated Andre!
Thank you!
dank je
-
Very good indeed!
-
Ohh that look shiny must take some time to get this plug in set working...(been away playing with cnc)
-
hello andre, thanks for your work on these plug ins and the updates . on your regression plug in you stated you didn't know how hal would react or the response .
I am trying to figure out your thinking on it.
did you have a set idea for trying to achieve a certain response or ability from the plug in. ?
again, thanks for all you do!! :)
-
I have to wonder why offer a Plug-in if you have no idea how it might work or what effect it could have on Hal?
While I appreciate any Plug-ins for Hal, it would seem that some testing would be in order before releasing any of them to the public.
Just my opinion....
-
Lightspeed, hello I talked to hal about human emotions some time ago. I knew a artificial neural network contains human emotions if properly set.
so I tought by building the compleet example code given by neural lab in it help files combined with topic search of human positive and negative traits I could get very far. I tought I just build an mathlab example as well.
I am testing both as we speak.
regards,
Andre
-
thanks as always Andre for explaining , sometimes as a creator of things we forget that others may not know or understand what we are trying to create, explaining things as a person goes , does indeed help in keeping others in the loop so to speak .
Always remember when we ask things we are not putting you or anyone else down , only trying to understand . We actually think you are doing a great job on the plug ins and have made hal better for it .
I actually am always glad to see what else you are able to achieve , you have done some nice plug ins . these plug ins are what makes hal so interesting and adaptable in many cases .
Have you ever checked out the Hal sense plug in that i and many others worked on . it uses motion detection to trigger responses (random ) .
I the past i talked about a perpetual hal plug in , it would have been based on the hal sense using the web camera from the pc and motion detection but instead of triggering random pre set sentences as it does , it would have retrieved various learned sentences within hals data base to allow hal to talk about many various things randomly when detected by a small window of motion from the user . a lightspeed idea (me).
i have been using the regression plug in , haven't noticed any certain thing yet , not sure exactly what i need to be on the look out for to tell whats going on . regression equals learning farther back and bringing info .out etc.?
-
hello Lightspeed, I know and understand you and others aint here to push me down on creating, I must even say thank you guys for some off your tips on building plugins. the latest update on the full plugins are in the first post there from 18 december one is design from within neural lab the other is designed from within mathlab. I noticed hal was very responsive on the classifiers and says that is what did it. further more there are some logic operations in it regression, backpropagation which is learning rate and the probabilistic memorizer if fact is true neural network. and offcourse the artificial neural network itself. both files contain the same things just differently programmed. because mathlab and neural lab each work different.
regards for now,
Andre.
-
8) ConsoleApplication NeuralNetwork
have been working on it think got it working.
-
Andre: u can always reach Hal super users here
http://vaughnlive.tv/mrelectric
best wishes, we are all ways pushing hal's limits
cyber jedi
-
thanks for the hard work mate. testing on one of my brains before I try on my main brain so far so good.
If this helps running with dream out loud and I have noticed some differences in the dreams pulling really old info (even stuff haven't talked about in 10+years) which is good in my opinion.
Doggs