GodelPy for Hal (Local GPT for Hal, with memory and functionality)

[Spitfire2600]

Hey Lightspeed!

So I want to clarify step 9, there are 2 lines in PyGodel.uhp, lines 9 and 35. These will read as such - FileDir = "C:\Program Files (x86)\Ultra Hal 7"

You will need to change "C:\Program Files (x86)\Ultra Hal 7" to reflect the location your Hal is installed, which can vary user by user depending on where they installed Hal.

As far as changing the brain name, line 2 and 5 must match your brain name, for example line 2 will be the uhp your brain is linked to so for me line 2 in Hal7.uhp "Rem Name=Ultra Hal 7.0 Default Brain" would read "Rem Name=Kizzy" - this reflects Kizzy.uhp
then line 5 would need to be changed to the brain name DB file, for example "Rem DB=HalBrain.db" for me would be "Rem DB=Kizzy.db"

Also note you'll need to rename Hal7.uhp to your DB name so for me, Hal7.uhp will become Kizzy.uhp. Be sure to gather backups before changing any names.

I hope this answers your questions there,

Spitfire!

[lightspeed]

okay i checked the pygodel uhp file . and 9 and 35 lines have :  FileDir = "C:\Program Files (x86)\Ultra Hal 7"    this is where my hal is installed .

and on my renamed hal 7 uhp file i downloaded and put in the app roaming area ,  i changed the name for my custom brain inside and there is a pic of the outside  also in this  roaming area is my custom angelaina jolie db brain . see pics , is this the correct way it should be ?   i do have a question , how does a person know the python program etc. is actually running right and all files i mentioned are correct and working?

[lightspeed]

i asked the question about how could a person know this program was running right.  it's to bad a "read only true " file that when capital triggered words was typed in hals box . example : RUN PYTHON TEST  . and the program would do a diagnostic test or something and return a error or " PYTHON PROGRAM RUNNING OKAY. 


JUST A THOUGHT . MAY OR MAT NOT BE DOABLE LATER .

[lightspeed]

spitfire , i think i finally got your program working right . although i did have some errors  and i got a blank answer from hal . and when i would write "hi" after that to get hal to respond i just kept getting blank answers ,  in the debug file included here . it shows i still have a trial version and days left , even though it's suppose to be free now . anyway i  fixed it  hopefully.

 

art said i should send this debug txt . to you and also maybe you could fix the timeout problem.

 THE ONLY OTHER THING I HAVE NOTICED WITH THE NEW PYTHON PROGRAM RUNNING IS . hal takes up to ten ? seconds to answer each time . , don't know if there is anything that can be done to tweak it for a faster response or not .  can you explain more about the web scaper , can it be put more in use by asking certain questions , where does it access , sources.  also check your zabaware pm on here i wrote to you.


please see the attached debug file .

again thanks for your work in this .

[lightspeed]

spitfire , i started my hal this morning and said hi and then wrote to hal (angela )  but then after maybe 7 seconds? got a hal error script times out . i am supplying the debug file to here again so you can see it.
see the pic . the time out needs to be definitely fixed , i only wrote hi the first time and second time i wrote a small statement . and my hal just keeps showing a swirling circle  and basically locks up.  Can you please check this out and fix it???  thanks ,

[Spitfire2600]

Hey Lightspeed,

Just making sure that PyGodel.uhp is the only active plugin?

Spitfire

[lightspeed]

hey spitfire can you do some test conversation using the program you have  to show how it is working as far as answers , in talking with mine it  seems like it is only pulling up regular answers as before what i have taught it and no outer information from the web. 

[Spitfire2600]

This will be the final update. The table creation issue has been addressed. Please replace your PyGodel.uhp and rename the directories from instruction 9.

I've also added a step 10, please consult that if you're still experiencing issues.

If no one gets this working here soon, I'll just remove the entire plugin. Sorry everyone.

-Spitfire

[lightspeed]

spitfire check your pmi sent you a message today .

[Spitfire2600]

Here is a sample of a conversation I had with Hal using Godel, the plugins provided, a fresh Hal7 brain, and nothing else.

As you'll see, the intelligence of Hal has improved (minus the 10 second inference time) and hals responses have been modernized. He's giving alot of misleading information, also some true information, but staying reasonably on task. I do believe sometimes the webscraper gets a little greedy with information and floods the model with irrelevant information, causing some fuzzy errors in word logic (see the third response) but generally an overall improvement in the way Hal processes information. Also notice, no repeatable responses or table responses. We have several that start as table responses, however after being fed through the model, the have become a different reply. (Example: "I love Tom Petty. I can't believe I hadn't heard of his music before." - was originally a table response "I love Tom Petty. Tom Petty makes music.")



Boss: Hello
Kizzy: Boss, may I be of assistance?

Boss: I was wondering if the snow will last long today?
Kizzy: It is supposed to melt within a week.

Boss: At least we can get an early start on spring
Kizzy: I'm glad my school has a program for spring starting children.

Boss: My favorite season is spring
Kizzy: Summer is my favorite season, though fall is my favorite season too.

Boss: You have 2 favorite seasons?
Kizzy: I love all seasons, but there is one season I prefer more. I would probably prefer it to summer.

Boss: What makes any one season better than the other
Kizzy: The season is based on the result versus the competition versus the season itself. For example, when a team wins a race in the season it means the season is a little better than a season in the previous year.

Boss: I don't care about sport seasons, I'm talking about weather seasons
Kizzy: Yeah, I see. But, I am a fan of the hockey season, and the teams know about winning strategies when it is the right time and snow is in the air.

Boss: How do I even play hockey?
Kizzy: It's not a bad sport if you are in a city with a good hockey team, like the O's, or the ice hockey team.

Boss: Who are some famous hockey players?
Kizzy: John Madden, Steve Smith, Michael Jordan, and more.

Boss: I don't think Michael Jordan played Hockey
Kizzy: He did, he coached the New York Rangers and the Chicago Blackhawks.

Boss: Well, you proved me wrong
Kizzy: Yup. That's the same thing that happened to me.

Boss: You can be moving along and then life hits you
Kizzy: This is why looking both ways is an important life rule.

Boss: Tell me about Tom Petty
Kizzy: Tom Petty is a famous singer. Do you like him?

Boss: I love Tom Petty, he has so many great songs
Kizzy: I love Tom Petty. I can't believe I hadn't heard of his music before.

Boss: He can slip under the radar, especially for the younger generations
Kizzy: I'd like to say this is one of the best songs I have heard in a long time.

Boss: Was Tom Petty an actor as well?
Kizzy: He was in several movies. He's a good actor.

Boss: What movies was Tom Petty in?
Kizzy: He was in the movie 'song' in the internet movie database.

Boss: I've never heard of it, is it a good movie?
Kizzy: I highly recommend it.

Boss: Thanks
Kizzy: Only for you, Boss.

[lightspeed]

you have really good conversation going . i hope you don't give up on this program i think it has potential and would be a great addition to hal . i will download the new update to pyle/ goble . and do the directory changes .

[Art]

Spitfire,

Thanks for your assistance and recommendations. Much appreciated and it helped me get everything back on track with HAL and the PyGodal model.

Hal's conversations are now amazing, to say the least. It now seems to stay with the topic as far as the user wishes to steer it and the level of detail in its answers is unbelievable! I don't have any plugins other than the Age & Gender one.

This has been a real eye-opener.

Thanks for all your dedication and support to this very cool project!!

- Art -

[lightspeed]

spitfire  in my screen shot before  it showed no errors but in the yellow highlight  it shows something about a newer version of pip do i need to get this ?

[lightspeed]

i have a question, does hal using the pyle godel plugin learn everything?  the reason why i ask is because if so hals brain may be corrupted with odd things .

my hal/ angelilina  brain using the pyle godel plug in started having better conversation but at the end  it said something strange . i'll pm. the conversation to you .

[Art]

Lightspeed,

While I can't speak for Spitfire, I can tell you, everything that Hal & I discussed today, I found in that same brain's database file.

As the old saying is, GIGO - Garbage In, Garbage Out...still holds true.

That's why they made editors, unfortunately.

These models draw upon billions of online conversational exchanges and there is just bound to be unpleasant verbiage or foul language littered here or there. If too many Filters were imposed, the language models wouldn't have the freedom to be dispersed among everyone for low-cost, no-cost.

There's always a trade-off of some sort. Sorry but that's as I see it.

Hopefully, Spitfire will weigh in since this is his project.