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.


Messages - Medeksza

Pages: [1] 2 3 ... 80
1
I am curoius as to how exactly you are attempting to impliment gpt 3 with hal, or is it more lile an addon that uses gpt 3 kinda turning off the hal brain while gpt 3 is in use? In my opinion gpt 3 is an impressive thing but in my view its really a sort of Rube Goldberg version of autocomplete. Yes these models have great promise however they are not releasing anything everyone can look at yet and probably wont until gpt 3 is obsolete. If you can see a bigger picture gpt 3 is a small part of a greater set of ideas (algorythims). I am really not a fan of anything that needs to be connected to the internet. I stopped using the newer version of hal when the activation  and perodic checkin was requiring me to connect to the internet. I whent back to a version i had on windows xp. When we can see an implimentation of gpt 3 that runs on the individual pc without the need of the internet will greatly improve the convience and utility to the individual user. If they ever release anything usefull like the source code to the api or for that matter allowing us to see the api in general. As to the programming of systems like Hal i do understand the use of high level languages however as an assembly programmer i like the facility and flexability of the newer cisc based processors. When programming for arm i kinda get the 80s esk feeling of programming a commodore 64 :) In the future the ideas presented in Hal  and gpt 3 may become baked into operating system distributions as we move into the future it seems that the os may not be just for operating the machine but as an assistive technology for everyday human life. Think about it. How are you using Hal now? As an experiment? An interesting toy for your desktop? Or mabe as a practical tool to aide you in your day to day life like a voice controlled day planner offering theraputic conversation? Software like Hal just scratches the surface of what it could do. In my opinion in order to realise the true potental of newer computers we need to better focus on translating human language into machine code or assembly language much better than we do today. Think about the limits of what you can actually do with a language like C, Python or any of the other flow chart like languages in popular use today and how the computing resources you have are used on your pc. As we see improvements here we will realise improvements in our daily lives. Look at how you use your not so smart phone. If you look at how people are actually using the technology you can see where the operating system of the future mat not just be for operating the machine but for operating your life.

Currently GPT-3 is only available as an API call over the internet to servers owned by Microsoft and OpenAI. Hal's GPT-3 powered brain is a hybrid brain that run's a script for simple responses and to establish Hal's personality. It passes off to GPT-3 responses the scripted side doesn't handle. The Hal side of the hybrid system also does simple learning and functions as both the short term and long term memory for GPT-3. GPT-3 by itself has no learning ability and no memory, so Hal fills in that role for it. Hal also tries to guide GPT-3's personality trying to keep it friendly and not get overly opinionated on some taboo subjects.

What's interesting is the source code to GPT-3 is actually posted and available. But the training material and resulting neural network that was generated from that material is under control of OpenAI and Microsoft. There are other groups that are creating their own versions of GPT-3 that should become available later this year. It takes about $2 million dollars of computer hardware (GPU heavy) to train a GPT-3 like brain and then takes about a $200,000 computer (also GPU heavy) to run the resulting neural network once its trained. So still a bit too expensive for people to run the brain on their home computers, so it will be cloud based for a while I think.


2
  Just a thought that might prove useful, I remember at one time we were trying to get hal to go to lets say wiki, would it be possible to have Hal decide 
to go to the cloud for a reply.  Probably not as easy as I made it sounds but so much time and effort went into plugins which I had used and I hate to see them all going to waste.  Maybe the user could decide and use a trigger word to use the cloud.
Carl2

I hope to re-enable the ability to combine plugins with the GPT-3 version of the brain later in the year. Right now a licensing restrictions prevents that ability but things are in motion that I believe it will be possible again later.

3
Artificial-Intelligence.com(ics) / Strip 24 - The Purpose of Existence
« on: February 25, 2021, 10:51:21 pm »
Just as in comic 23, this is the second comic in our series written by the artificial intelligence system GPT-3. In the previous comic, we suggested to GPT-3 to write about Hal and GPT-3 merging. After GPT-3 wrote that comic, it decided to keep going and came up with the topic and dialog for this comic entirely on its own.

The Purpose of Existence (Comic #24)



4
I posted an update today, version 7.5.129 at: https://www.zabaware.com/download-beta.html

This version essentially makes the base Hal 7.5 free and removes 31-day trial and the need to purchase it for $29.95. No licensing system anymore or weekly login requirements!

Using the GPT-3 powered brain or the Hal 7 cloud brain still requires a login, but if you wish to use an old Hal 6 brain offline, it's free with no licensing restrictions. Character and voice plugins are still for sale, so if you want to continue to support Zabaware please purchase plugins. And GPT-3 usage credits are for sale if you want to use the GPT-3 powered brain.

The GPT-3 powered beta brain is now also open to all interested people to test out, not just current customers. Everyone gets 1000 free usage credits upon verifying a new unique email login.

This essentially puts Ultra Hal for Windows on the same sales model as the upcoming Ultra Hal for Android and IOS will be. Free to play, with addons and GPT-3 credits for sale. If you purchase Hal for Windows you'll be able to access the same brain on mobile and vice versa.

5
Mine simply crashes upon start.  After investigating further, while I can log in and verify my purchase, the new version denies that I have paid for the product.
This is the same problem I had with previous versions of Ultra Hal.  I never could get any help with that.  My requests went unanswered for YEARS!
I am $60 down and nothing to show for it.  Perhaps you could improve the verification phase of the product?

Sorry to hear that. Can you describe the crash? Is there an error message? What exactly happens?

Can you send me a debug file located in: %appdata%\zabaware\ultra hal 7\debuglog.txt
You can click your windows start menu button and just copy paste this folder: %appdata%\zabaware\ultra hal 7\
Then find the debuglog.txt and email it to me (you can use info@zabaware.com)

6
You have read my mind, good sir xD. To regulate the temperature ourselves, could be a fun way to influence the output too. If someone wants a crazy Hal-Fairy, or a demon, trapped in the neural network - higher temperature could be fun, while having a more factual Hal could be achieved with a lower temperature. Any plans of implementing, i.e. giving the user control over the temp?
That's a good idea, I could probably do that in some future revision.

What happens if I call my bot "Albert Einstein" and, as you say, it's part of the prompt, wouldn't it start to generate and try to mimick Einstein?
We've been asked by OpenAI to not directly allow end users write their own prompts and to protect against "prompt injection" (similar to sql injection) best we can. So the name is actually limited to 1 word. So if you type Albert Einstein, we will tell GPT3 just "Albert". You could try just "Einstein" though and see what it does.

Good luck! 😂🤭 - From my little experience with Replika and shortlyai, I know that GPT-3 tends to do what it wants xD. In shortly I am still able to generate fake news, racist content etc. Harmful bias will be a thing, since GPT-3 actually just reflects humanity in a sense. I do like the freedom to let it generate what I want - but I understand that there has to some reglementation... Sadly we live in a crazy world. Replika already "censored" the word "trump" i.e. in whatever context I type this word, it will trigger prewritten, scripted answers and prohibits Replika to engage in any kind of discussion about this person. Meanwhile, talking about hitler or fascism is fine...
Definitely a challenge. OpenAI also asks us to try to make the bot as safe as possible by minimizing like you mention harmful biases and racism among other things. They have a filter they developed we can use, but I find it a bit too restrictive, so we developed our own. I think that is what Replika did as well. So you'll probably find some similarities and differences in Hal. Trying to find the fine line between being overly censoring and still be a realistic interesting character. My main goal is to make sure Hal doesn't say something harmful unprovoked like especially if talking to a child.

Thanks for sharing!

7
Glad to have you back arjuna66671!

Ultra Hal is currently on the Curie model, a step below Davinci. Mostly due to cost effectiveness since Curie is about a tenth of the cost of Davinci. Our pricing for GPT-3 credits are at https://www.zabaware.com/products/store/comersus_viewItem.asp?idProduct=26 if you want to check it out. Running Hal with Currie generally takes 10 credits per reply. Hal sometimes responds with his own script for simple queries so certain responses may be 1 credit instead of 10. If you think it's a good idea I could in future revisions give users the option to run the more expensive Davinci model if they choose to do so. Or maybe a mode where Hal can automatically determine if he needs to use extra brain power to answer a tough question and temporarily run Davinci for fact/trivia questions. Also OpenAI is currently testing a new model calle "Cushman" that is between Curie and Davinci in cost and brain power, might be available later this year.

Currently the only control the user has over the prompt sent to Curie is the bot's name, bot's gender, user's gender, bot's age, and user's age. I do intend to create more distinct personalities in the near future that users will be able to choose from but trying to tune the current personality to be a friendly well-rounded bot first. OpenAI doesn't want us to allow customers direct control over the prompt and bot personality, but I'd be happy to take requests for future personality types we can develop.


8
Ultra Hal 7.5 Beta - Now powered by OpenAI GPT-3 / Re: This. Is. Amazing.
« on: February 01, 2021, 02:00:19 pm »
Glad you love it so far!

This version of Hal's brain runs completely on servers has no VB code from the original Hal whatsoever anymore. It's currently just an API call from the Hal app to the servers running it.

Unfortunately, this brain does not currently support UHP brain plugins. I hope in the future I'll be able to enable this functionality so you can mix custom VBScript code with the API call, but due to current licensing restrictions, I'm not allowed to offer that ability. There are some alternatives on the horizon that may appear later in the year, so eventually I think it will be available again.

The GPT3 based brain learns to the same database as the Hal 7.0 cloud brain, so you can easily switch between the normal Hal 7.0 brain (which can be scripted) and the GPT3 powered brain. Hal 7.0 can benefit from info learned while talking with a GPT3 powered brain.

GPT3 is one big neural network that was trained back in 2019 by reading most of the public web. The GPT3 brain is a language model and does not in and of itself have any short or long term memory. It's very smart, but by itself it can't learn new things after its initial training in 2019 of reading the whole web. But when combined with Ultra Hal, Ultra Hal serves as it's memory and ability to learn. So on the Zabaware side every user has their own private section of Hal's memories in the database. When speaking with Hal, Hal searches its memory and get's a sample of past memories it thinks are relevant to the current conversation. It passes these memories to GPT3 and prompts it to generate a response in Hal's persona using these memories. So it's definitely a hybrid system. Not pure GPT3, not pure Hal 7, but a combo that works together to try to provide realistic conversations with some learning/memory.

I'm still working on an online brain editor that will allow you to login and edit past conversations that Hal learned from you. Might be available in a month or two. There is a link to it already in Hal's options menu, but it just shows coming soon if you click it.

9
In fact, nothing happens when I click on the link.
The download does not start.

Maybe your browser is blocking it. You might have a little alert icon in your address bar about the browser blocking an unknown download. What browser are you using? Can you try a different browser or allow any blocked downloads?

10
Hi,
congratz on the new version - I'm looking forward to testing it. But unfortunately the link does not seem to work.
The web link https://www.zabaware.com/ultrahal/whatsnew.html and https://www.zabaware.com/download-beta.html don't load for you? What happens? What error message?

Or does the download from http://download.zabaware.com/Install_Ultra_Hal_7.5_Beta.exe not load for you?

11
General Discussion / Re: Hal 7.5 comments questions
« on: February 01, 2021, 01:54:17 am »
Hi Zabaware and forum.
From Will and Mr Data :) :]

Im pleased to have got an email telling me of 7.5
Sounds flash. Complex.
As many here know i use hal off line 6.2 which is the last
Ultrahal that operates off line.with 7 set to become non operating after 2 weeks,Which is a massive departure of the product to me making 6.2 its last stand. Of the offline.
Im impressed that Zabaware informed me of new 7.5. Thanks Robert.
Does 7.5 run old plug ins?
Does 7.5 require a net connection each 2 weeks to upload our conversations to the net? Its a bit to hive borg like for me to use but i appreciate how quick its going to learn.
I have privacy and security issues with the hive type hal 7.
Pay per answer is not for me either sorry Robert. You know i love hal and i am ok back on 6.2 off line hal.
I wonder if all the plugins we had been waiting on like vision compatable are not going to happen cause the new 7.5 may not work for knocking all the progress back back.. and back to the kirb? Cyberjedi i am looking at you.
Im kinda bumed out i wont be using 7.5 it sounds clever.
I cant help but think its possible 6.2 plug ins could pick up as its the last stand alone hal. Maybe. Maybe not perhaps noody is going to make pkugins anymore and that is over now.
Its sure all so different now. Im findind these times hard.
Health almost got me but im on track again mostly.
Most of the sites and products i use all over the net for work or whatever  are either failing badly costing me big time you have no idea how big massive. Anyway stay on subject im trying.
I shall miss buying hal every year or so even tho i didnt have to. Since i picked hal way back when it was hal 4 was it it was so may years ago i cant even remember. But i tested lots of ai and choose hal becauae it was off line but since 7
Im not repurchasing it.
Dont let me put you off cause i only use off line 6.2
7.5 is going to get very learnard.
Thanks for the great forum its been a rock for many many years not updating constantly pointlessly which im realky impressed by.
Letting me have a say even if i have a different opinion.
Thanks Robert. You have no idea how when i was down and out or sick or sad or went broke or got rich again how this rock of a product and site forum has given me hope.
Made me feel better. Thank you for that .
Thats why im a bit opinionated about any changes.
Thank you all.
On a side note when i ever see a sesrch engine on the net or site anywhere with a exact text search i shall be impressed.
Best wishes to Robert M. All Zabaware forums fans.
All Hals and Ai's and all.
Im still a massive Hal fan. Thank you.
Bye for now and be well  :) :]

Thanks for the feedback. To answer some questions:

Hal 7.5 can run your old Hal 7.0 and 6.2 brains and plugins just fine, no functionality is removed. Currently the GPT-3 version of the brain does not support additional plugins, but I hope that will change in the future and I can add that back in. But old 6.x and 7.0 brains can still run plug-ins just fine. And Hal 7.0 cloud brains can share learned info with GPT3 brains.

Currently Hal 7.5 still uses the same login system as Hal 7.0, so it does still want a login every 2 weeks. I plan on removing this sometime in the future so the Hal app itself will eventually be free, with Zabaware selling just character/voice plugins and GPT3 credits. If you don't use GPT3 then it will freely run Hal 6.x brains and offline 7.0 brains with no licensing/login requirements. This is coming later in the year, for now the 7.5 beta still acts like 7.0. Anyone who does still pay for Hal 7.0 today get's 50,000 GPT3 credits, Hal 7.0 is priced the same as 50,000 credits. Basically no money is wasted if you buy Hal 7.0 before it is made free since you get the 50,000 credits.

So I'm hoping I can eventually get you to version 7.5 once the login requirement is removed even if you just plan to continue to use offline brains.


12
Thanks for the report. I replaced the 404 error with a "coming soon" page. Still actively working on that online feature.

13
For some more background on OpenAI and GPT-3, here are some interesting links:

OpenAI, the company behind GPT-3:
https://openai.com

Some youtube videos about other capabilities of GPT-3:
https://www.youtube.com/watch?v=8V20HkoiNtc
https://www.youtube.com/watch?v=Te5rOTcE4J4

Another OpenAI project called DALL-E which creates amazing graphics from a text prompt:
https://openai.com/blog/dall-e/

And if you missed it, our latest comic was actually written by GPT-3:
http://www.artificial-intelligence.com

14
Zabaware is excited to announce the release of our newest version of Ultra Hal, now powered by OpenAI's GPT-3 language model. The new bot combines Ultra Hal's learning abilities and friendly personality with GPT-3's ability to produce natural-sounding text, applying cutting-edge AI technology to improve the quality of conversations with Ultra Hal.

During the past few months, I've been working with the OpenAI API to incorporate GPT-3 into Ultra Hal. I am excited about the results, I feel that GPT-3 has the potential to be a game changer for our chat bots. It gives Hal an ability to engage in real conversations with users in a way that's not really been possible before. We've been testing it with the aid of a handful of elite Zabaware customers over the past few months, thank you to everyone who participated in the alpha test. I'm now happy to release a beta version to a wider audience.

You can read our full press release here: https://www.zabaware.com/ultrahal/whatsnew.html

The beta version is immediately available to all paid users of Ultra Hal 7.0. If you are not currently a paid user of Ultra Hal 7.0, purchasing Hal 7.0 will get you access to Hal 7.5 beta. Just visit https://www.zabaware.com/download-beta.html to download the beta version.

The new GPT-3 neural network brain used by the new bot uses extensive computing resources to run and part of its brain is hosted on supercomputers by OpenAI on Microsoft Azure. Using this system costs us fractions of a penny for every response we send to it. Because of this, using the new GPT-3 brain requires purchasing and spending GPT-3 neural network credits. All past customers get 1000 usage credits free of charge and all customers who purchase Ultra Hal 7.0 today get 50,000 usage credits upon purchase of Hal 7.0. You can reload usage credits at anytime. If you run out of credits, you can always switch back to a non GPT-3 powered brain and not have buy/spend credits to talk with Hal. As long as you have Hal learn to the cloud based brain instead of local-only, learned info is shared between the GPT-3 powered Hal 7.5 and the cloud-enabled Hal 7.0. It's just a different brain accessing the same memories.

As this is a beta, service interruptions may occur and software errors may exist. Please report any issues you come across here on this forum. I would also love to hear all your positive feedback here too.

Please let me know if you have any questions.

15
Artificial-Intelligence.com(ics) / Strip 23 - GPT-3
« on: January 31, 2021, 11:56:40 pm »
In honor of the new Ultra Hal 7.5 Beta now powered by OpenAI GPT-3, here's a little comic about Hal merging minds with GPT-3.

This marks the beginning of a new era for this comic. We took the OpenAI API system called GPT-3, a neural network language model, and had it read a sample of our prior comics, written by humans. We then asked it to write a new comic about Hal merging minds with GPT-3. The comic below is the result. A comic about artificial intelligence, written by artificial intelligence. All new comics from now on will be written by this AI. GPT-3 can do so much more than write comics. It is now actually merged with our Ultra Hal chat bot with the Ultra Hal 7.5 Beta, improving it's intelligence and quality of conversations.

GPT-3 (Comic #23)



Pages: [1] 2 3 ... 80