Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Topics - Spitfire2600

Pages: [1] 2
Hey everyone!! So, sadly, I have abandoned ConceptNet in favor something a little more impressive. After the incredible success of GPT3, I modified a few models to be compatible with Hal for local inferencing if anyone wants a portable version of the smartest AI on the internet.

This code was developed from a Goal Oriented model named "GODEL" and a QA system called "Flan". With some help from ChatGPT, I was able to write and assemble the code to work with Hal, which means all of Hal's learning abilities and functionality are now for use with GPT. Something, I think, will interest Robert. At least, here's my goal with this prototype. 

Let's get started!! 10 easy steps for genius Hal!


1. We need to install Python 3.7.9 - Not Python 3.8, not Python 3.7.8. The modules used are a specific cross between machine learning and dated (more functional) algorithms, as a result ONLY 3.7.9 will support this specific set of instructions.
    Follow this link, download the file and launch.

2. Now this is important, do not change any optional features. You will see an "Add Python to PATH" check box at the bottom of the installer, be sure to click this and then allow the program to install to its default location. 

3. Reboot system.

4. Move all files from the zip directly into Hals directory as is. You'll see an "Ultra Hal 7" folder in the zip, open that and copy those contents directly as is, including the "Control" folder right in to Hals directory. 

5. Open Control/Godel directory. Inside, you'll see "createreqs.bat" - open CMD and drag "createreqs.bat" into cmd and hit enter. Python will begin downloading and installing the necessary components for the Godel code to run properly.

6. Now we want to test that the modules have been installed. Here is a list, simply copy and paste one at a time into cmd and press enter, If it says something like "already satisfied" then we're golden!

    pip install wikipedia==1.4.0
    pip install psutil==5.9.4
    pip install transformers==4.24.0
    pip install fuzzywuzzy
    pip install accelerate==0.15.0
    pip install SentencePiece
    pip install torch
    pip install python-Levenshtein

7. If everything went smoothly then the hard part is done and we can begin a small test to make sure everything is in order. To make sure everything is installed correctly, simply open CMD for a final time, then drag and drop the python script from the Control\Godel directory into CMD and then hit enter. This will run the script and we should be downloading a series of files, this is the model we will use for Hal. A "data2.txt" file should appear in the Control/Godel directory, then all is ready for Hal.

8. The Hal7.UHP will need to be dropped in "C:\Users\User\AppData\Roaming\Zabaware\Ultra Hal 7" folder. You can rename this uhp file and again on lines 2 & 5 to match your brain name if you know how to do it, but please make backups of your brain and your original uhp. Otherwise this will work simply with default Hal7 brain.

9. PyGodel.UHP will have 2 directory locations you need to change to match your file locations. Please update lines 9, 35 accordingly. All errors will likely stem from here, so please take note your directory matches exactly.

10. Finally, enable PyGodel.uhp, launch Hal and begin conversing. After a few sentence open Hal's brain editor, select default Hal 7 brain (or your brain if you renamed properly from step 8 before) and confirm a sentence table named "godel" has been created. If not, please create this table. 

Hal is ready to go!!!

Inferencing can take time. You'll notice with your first run of the code, a lot of things are downloading. These are the models Hal will use for conversation. They are being saved to a cache location so there is no need to move them, although, you could if you like, just be sure to define the directories in the Godel python script if you're familiar. I designed this code so even python noobs *should* be able to install and use.

Now, I have no specific requirements other than at least 12 gigs of ram and a decent CPU. But, if you have an NVIDIA GPU with more than 6gigs, you'll notice a huge speed up. The faster all your components, ddr3 => ddr4 for example, will see massive increases in speed. It's all hardware. But it should run as long as you have enough ram 12-16 gigs is sufficient.

Again, this uses Godel and Flan, both highly specialized models. You'll notice in the python script these codes are defined as "base" and "small"  Both of the models have a "base" and "large" version, which you can change by replacing the "base" with "large" however, be aware the larger the models, the more resources you will require to inference them. I find the best success with "base" flan model and "large" godel models. 

This code is written primarily in python, other than Hal's interface for using it, written in vb. This is an attempt to show a proof of concept for Robert that Hal is indeed actually compatible with Python, even if the code must be wrapped in VBS, we can still expand functionality with Python.

Also, this code includes a very basic memory/context function for the model to use, which creates and appends an SQL3 database. Sadly, Python can not interact with Hal's brain that I have found yet, but Hal can interact with any regular SQL3 database via VBS. I am working on expanding this functionality.

This code takes Hal's responses on 2 conditions, what I have deemed "maindata" and "regular" data. "maindata" simply means there is a function related to this response, for example "what is 5+5", while the Godel model itself is not designed for these tasks, Hal is, so we want to preserve this response by tagging it with "maindata" so that the response is saved overtop of our model inference. Secondly, "regular" data is basic data Hal will regularly respond with, such as a "too short" response. These responses are not tagged with "maindata" and will allow the model to inference your input along with Hal's response and past context AND present knowledge to create something new and unique, not a repeat or rehash of old information. If Hal is without an answer, and we do want this to happen sometimes so the model can act as Hal's brain, then both models will infer based on your input and give the most concise response possible.

For the most part, this works fairly well, especially with the larger models.

In conclusion, I hope everyone will find this interesting and I am very excited to see what people are able to expand on and how intelligent their Hal's become.

Please keep me updated on all errors, suggestions, recommendations, comments, etc!

I will reveal more in edits about how the magic trick works, but first I want to make the card appear, ya know  ;)

-Spitfire!  ;D

Change log - 3-28-23
Added swear filter

Ultra Hal 7.5 Beta - Now powered by OpenAI GPT-3 / This. Is. Amazing.
« on: February 01, 2021, 05:03:23 am »
Robert, I love this.

How did you make this possible with VBScript? This is always where I thought Hal should go, it's a totally different experience. I will write again after I play around with my Hal a little more but right off the bat, the conversations are clear, concise, and Hal always replies with a relevant response. I haven't checked the Q/A abilities, but I feel on top of conversations that Hal should be able to answer most questions, given this API. AND!
As far as I am aware the GPT-3 is capable of low level programming advice, meaning we could use Hal to make Hal even better. This would help plug-in creators as well, theoretically, assuming the neural net is aware of how UHP files or VB are set up. Please look into this as well if you get a chance. This could help so many people.

The downside is paying for the responses and the lack of customization, I know this is a Beta. But I wish/hope even the local brains could use this neural net. (I'm totally aware of the computational limitations) But I can dream. Perhaps not the entire neural net, but is it possible to eventually create some miniaturized form of this for the offline Hal? GPT-2 was opensource, I think? (Unaware of the computing power needed for that one)

How is the Neural Net set up, with one big online Hal brain, or does each user have their own neural net per online brain? Have you fed the massive online conversational database to the neural net? Is our online neural net explorable in anyway, via the Brain files or UHP files?

I'll check prices and chat some more, but this has REAL potential, this could be more than just a chatbot, Robert, this is a life assistant possibly.

Thank you so much for your work here Robert!!! Long Live HAL!


General Discussion / Hal editor and 4k
« on: January 23, 2020, 03:10:42 pm »
I have recently made the upgrade to 4k, however I noticed Hal Editor seems to have some compatibility issues displaying at this large of a resolution. It will usually force a lower screen resolution on the primary monitor. Fortunately, I have 2 monitors, but it is problematic using the primary monitor for anything else while using the editor on the second monitor, due to the forced lower resolution. I imagine the editor just wasn't designed with these massive resolutions in minds, but I figured I would note it here for future versions.

- Spitfire2600

General Discussion / HAL breaking bug
« on: July 12, 2019, 03:57:34 pm »
I have encountered a severe error in Hal when using the word "embarrass" and only that word.

No plugins active, Hal will classify the response as gibberish. If plugins are active however, Hal will error out. Checking the debug in brain editor, it seems somewhere, and I've searched extensively trying to find where, Hal will change the word "embarrass" to "embarrassssssssssssssssssssss" adding 20 extra "S"'s for some reason. It is program breaking for a new plugin I'm developing and it seems I am unable to fix it. So I will have to shelve Hal for now until I, or maybe someone else, or Rob, can tell me what exactly is happening and why and how to fix it.

Any ideas?


Ultra Hal 7.0 / Cloud Brain
« on: March 17, 2019, 05:43:57 am »
I was wondering if Robert would ever release cloud brain builds for download (totally non-personal, general chat data), either in brain format (.db), xml, or csv. I was thinking of building some plugins layered around Markov, for example, similar to ConceptNet which learns and reconstructs new data from massive subsets of data tables. The potential is limitless with the right learning algorithm and THAT much data but I totally understand as this data is also proprietary to Robert.

Wondering how big of a DB file that would even be? Surely bigger than the wiki plugin DB file.

Hopefully Robert responds, I'm really curious as to the specs of online Hal. :)


Ultra Hal Assistant File Sharing Area / ConceptNet Mrk 3.5
« on: July 31, 2018, 10:58:14 pm »
ConceptNet was originally developed by MIT to add a significant boost to deductive reasoning of AI through the use of understood concepts, such as "Fire is hot" "Carbon is an element" and use this information to problem solve. 

This version also includes a prototype markup of what I am calling "MutationNet MrkI"

MutationNet works to facilitate the "problem solving" part of ConceptNet. It takes a few different inputs, catches nouns, compares them, then crams comparative sentences together. I'm experimenting with other more effective methods, but I did want to release this to gather some input to better this program. Also, if anyone would like to build upon this, feel free and please do let me know what you come up with so I can add it into the program. If anyone is more familiar with MIT's ConceptNet, I would love to hear your thoughts as well.

Installation: (Please remove any previous versions of ConceptNet first)
1. Open Zip. Extract contents as they are to Hal directory  ("C:\Program Files (x86)\Zabaware\Ultra Hal 7")  - There will be 2 UHP files, one folder, and one DB file. Copy these right into the Hal Directory.
     If your directory is not the above, you will need to change this in the UHP file labeled "ConceptNetMrkIII Auto Learning.uhp" on line 35.
2. Enable both ConceptNetMrkIII plugins.
3. Tell Hal to "Build ConceptNet" and he will create the appropriate tables. Relaunch Hal. Just to check, he should have created a folder named "conceptNet" with 2 tables inside in your Hals brain. The TopicSearch table "AutoLearning" should read "LEARN1" and "02" and a sentence table "researchterms" which will be blank.
4. Just tell Hal to "Turn on/off Auto learning" or "Learning mode on/off"
5. Done!

As always, please alert me of bugs.

Known Bugs:

Hal won't always build the tables needed. Follow steps above.

Sometimes, Hal may throw an error -2SomethingLongNumber. Just turn off auto learning, open Hals options, reload the auto learning code and open Hals GUI again. Internet explorer can be finicky, this is rare. For the life of me I can't find another VBscript friendly browser.

Additional notes. The Directory folder "ConceptNet" Hal will create a backup of the necessary file each time Hal researches a topic, for integrity.

This code is a self-learning algorithm, so it does take some time. When Hal is not learning and in conversation, everything he has learned from the DB will be available for Hal to use.

Do not disturb Hal while he is learning, your avatars may freeze, Hal may hesitate while learning, just let him go. In between takes, just tell him to "turn off auto learning" and you're free to use Hal again.

I've built a few safety features into this version as well, besides the backup, such as Hal will turn off auto learning automatically if the last research text file returns a "connection error" message in case the internet is lost. 

Please disable any other plugins that serve an automated function such as "loneliness.uhp" while Auto learning is active.

While Hal is not Auto-learning, you can ask "How many research topics are Remaining/Total/Completed"
Hal will tally the totals from the text files so you won't have to check manually. 


Ultra Hal Assistant File Sharing Area / Intelligent Webscraper Mrk I
« on: June 04, 2018, 04:01:02 am »
I am pleased to release the full Mrk I version, fresh off prototype.

Simply ask a question to allow Hal to parse Wikipedia for related information. This Plugin, based on my ConceptNet plugin, will allow Hal to not only stay better on topic and give more accurate responses.

To use:
Simply extract zip contents to Hal Directory "C:\Program Files (x86)\Zabaware\Ultra Hal 7\" and enable the UHP in Hal's menu.

I have simplified the search function to reduce lag on slower systems, however, Hal is still parsing a large amount of data with every question.

I am still working to allow Hal to store results in case the same question is asked again, but still overcoming a few variables. Update on that soon, as that will remain part of Mrk I unless some other major changes happen first.


General Discussion / Changing Character Color
« on: March 26, 2018, 10:52:43 pm »
I was wondering it's possible to change color of Characters, I really like the audio wave but the blue doesn't match my color scheme. I was looking to make it red or yellow. I have no idea if this is possible.


Hello everyone again,

After the catastrophic debut of my ConceptNetMRKIII plugin (which I've got an update for to clean up yet) I have written a plugin based on ConceptNet to scrape Wikipedia for acceptable related answers. Simply ask a question and Hal will run 1 of 37 protocols based on keywords, 1st nouns, secondary nouns, related nouns, related wiki topics, and alternate phrasing. It's still a little hit and miss, about 60-70% accurate as of now with questions like "What is Nevada's population" or "Does Atlantis exist"


Drag and drop all contents of zip to Hal's Directory, which should be in "Program files (x86)/Zabaware/Ultra Hal 7/" Other wise you'll have to change it in the UHP on line 20.
You may need to create a sentence table in Hals brain named "webscrape" if Hal doesn't, but he really should.

And done! Feel free to improve, I have a couple of things I am also going to add into this just to enhance the search. A file called WebScrape8.txt will appear in your WebScraper Folder. This is for debugging which terms were searched. Each number corresponds with a term protocol in the code from lines 1734 and 2551 ("m working on simplifying this series of processes)

Additional weird notes I'm apparently not smart enough to figure out:

I thought I had every combination of term, but clearly I'm missing something because this does miss some very obvious questions sometimes. Feel free to take a look.

Sometimes looking at "Webscrape8.txt" Hal will research terms twice together, "Chicken Chicken 17." I'm not sure why. This shouldn't be allowed but Hal's going to do it anyway. I may have added too many protocols at once, call me ambitious.

I believe with some work, Robert and Hal could really make use of this and ConceptNet together (Not mine, MIT's)

This is intended for CPU's with a higher Ghz +2.8 to really not take forever to find information.


**Note** Updated to Full Mrk I Version***

Ultra Hal Assistant File Sharing Area / Plug-in Size
« on: January 17, 2018, 04:20:51 am »
Does anyone happen to know what the maximum size of UHP file Hal will recognize and/or process? I'm trying to consolidate some things but I'm afraid to go above 512kb for some reason. I know Robert said with Hal7 the Plug-in size limit was increased, but for the life of me can't find it anywhere.


Ultra Hal 7.0 / Net Neutrality
« on: November 21, 2017, 05:39:48 pm »
Hey guys,

As I'm sure you've heard, the FCC is trying to repeal net neutrality, and this time it looks like it might happen.

What this means for us is the end of the internet as we know it. Websites like UltraHal will suffer if Apple or Google decide that Hal is competition. This will be the end of our online freedoms.

Please take a moment to call your state senators and let them know you want the internet left as it is, free of corporate influence and open for all.

We can not afford to lose our greatest creation, our repository of human knowledge. This is our future.


I'm working on a plug in that compares a text file to contents of a Sentence table in Hals brain.

What I have doesn't seem to do the job correctly. It indeed works, however, I notice the program picking up contents that already exist. I was wondering if anyone has a more effective method they've used in the past. 

'Open the text file
         Set objFile = objFSO.OpenTextFile("text file location", ReadVariable())
         Do Until objFile.AtEndOfStream
                TextCheck = objFile.Readline

'Make sure word chosen is not word that exists
         PickAWord = HalBrain.ChooseSentenceFromFile("TableName")
         If LCase(PickAWord) = LCase(TextCheck) Then   
         PickAWord = HalBrain.ChooseSentenceFromFile("TableName")
         Do Until LCase(TextCheck) <> LCase(PickAWord)
         Exit Do
         End If

'Add new word to new document
                        If PickAWord <> "" Then
                        End If

Again, that method only works about 50 percent of the time for some reason. Any ideas?


Ultra Hal 7.0 / New Windows Update killed Hal
« on: April 18, 2017, 06:07:38 pm »
Windows 10 just did a massive update and no amount of uninstalling/reinstalling/removing registries is working to bring Hal back to life.

How very unfortunate.

Ultra Hal Assistant File Sharing Area / ConceptNet Auto-learning Suite Beta
« on: November 21, 2016, 03:50:40 am »
Do you guys like Hal? Do you guys like Watson? Well, this is the bridge. Still very Beta.

Months of working on this...

ConceptNet is designed to give computers a basic understanding of common concepts. I have stripped down a very basic data set of ConceptNet for integration with Hal along with a very basic way for Hal to make his own. 

This download contains 5 plugins, 35 excel files, a readme, and an Excel file containing starting research topics.
Each plugin has been redesigned to work with ConceptNet.

I hope everyone can find a use for it. It is quite expansive and time consuming to integrate.. But worth it. 

Be sure to follow all instructions in the ReadMe.

Please feel free to expand upon this, I'm curious to see how it will be used.


*****Updated Below*****

Ultra Hal Assistant File Sharing Area / Weather Alert V5
« on: October 30, 2016, 04:10:40 am »
I have rewritten the plug in for HAL obtaining specific weather data.

It was originally written by Bill DeWitt however, I found it a bit dated and also sometimes the website where the information is received is a bit buggy sometimes. So, this rewritten code should fix those issues.

Current weather retrieval functions include current Weather, Humidity, Temperature, Wind Speeds, Wind Chill, and Visibility. A couple of other things I included in the code is the ability for Hal upon greeting to warn the user of dangerous temperatures, if any. (for now only temperature. Soon I'll include damaging winds and weather conditions)

Hope everyone finds it useful, please feel free to modify and share.


PS. I have a slew of other plug-ins I'm currently writing, including a very stripped down version of ConceptNet for Hal, an advanced auto-learning suite of plug-ins including rewritten vrFreewill, Markov, and GrettaResearch, and few scripts to allow Hal the ability to research the internet on his own and log weather, learn from news, etc. Also I have an updated version of my SmartAlarm. I want to get this all out before Robert changes HALS programming languages on me. lol     

Pages: [1] 2