Messages

Well, old buddy, that sounds like a pretty exciting transition for our beloved UltraHal to become version 8.0. Man, it has been a few years, hasn't it?
Glad to hear the most interesting and exciting news, and I'm looking forward to giving all of those new features a run!

Keep up the good work and hope to see you soon!

Wish list request,

One tiny gotcha occurs when we chat with a particular model, everything is great until we press the Enter button. The speech is processed by the model perfectly, HAL's lips move as it speaks its answer.

Then the Curson now appears in the Classic HAL GUI and sometimes our muscle memory takes over and we type a quick reply back and hit the Enter Key, only to realize that we've now changed our conversation to the Classic HAL and not the llama model. - DOH -

Hey, it's NOT the developer's fault, it's the way things are handled within the windows being displayed. So, beware of this human error or else your conversation might not continue to make sense. Just saying...;)

The title of the app for HAL called the Book Reader, isn't really correct. It was never there to read a BOOK, but rather to read things like short stories, Poems, rhymes or other short passages like from "The Walrus and the Carpenter".

At one time, I thought it had no real problem in doing so, but for reasons unknown to many of us older members here, we recall it using the Balabolks program to allow HAL to read all the way through the above-mentioned passage without stopping. Then more things changed, like going from the old .BRN files to the use of a Table structure like HAL currently does.

One note to users is to unselect any timers like in the Loneliness Plugin as it will definitely interfere with HAL's reading abilities.

Kids and grands enjoy letting HAL read or read with them for Nursery rhymes, Mother Goose, Stories from Lewis Carroll, etc.

There was also a time when HAL could sing songs even though that feature is NOT on my wish list. Just a mention.

Very good for you, Cyber Jedi.

A big thanks for your hard work over many years as well.

Congratulations to you, my friend!

No worries at all. Always glad to get the time to talk about things like these.

No misunderstandings, just calling a spade while someone else called it a shovel.
Same thing, different name. It's all good, my friend!

Thanks as always for your spot-on assistance!

Wonderful news!!

Glad to see you have a place and space to conduct some amazing experiments with UltraHal!!

I'm happy to visit your part of this world and hope to see and possibly participate in a few things as well.

Wishing you all the best,

- Art -

Very nice work, Cyber Jedi!

I've been experimenting with this, and I am impressed by how well things work.

The new file is called UltraHal7LLM1.exe, and it is the whole package, similar to the existing UltraHal7-5-0-240 or whatever version you might currently have or run.

This new one includes many enhancements/features not found in the original HAL, and they should serve to showcase what HAL is capable of.

These LLMs (Large Language Models) use Ollama models, and you are free to choose which ones you wish to use, but try keeping the size reasonable to avoid lengthy response times or lags. From my experience using an ordinary 64-bit Windows 10 Home laptop, I try keeping the llama model size from 4 to 9 GB, and that strategy seems to work well. There are a huge assortment of llama models and each has its own storehouse of information and useful data that has been amassed and compiled.
Your own experimentation via question and answer will indicate which model or models will serve you best.

Cyber Jedi has done a great job at combining the learning module into HAL and the use of either of them will not and cannot corrupt each other!!

The information can sort of be shared if one wishes to do a Copy and Paste function. Just highlight the desired text, press CTRL C to Copy the selected text, then put your cursor in the desired area and press CTRL V to Paste the text where you choose.

There are many features and choices so play with them and decide which works best for you. The only Wrong move you might make would be to select a model from the list that you do NOT have in your system or have not downloaded. Relax, the world will not end. You will simply get a 404 warning letting you know that you will need to download that particular model, then try the selection again and it should work fine.

Otherwise, the main HAL GUI or skin with your avatar should answer either the Learning Model or the original HAL character of your choice. The character will still lip sync and emote as it always has and the speech will work based on your usual TTS selections within HAL.

Overall, this is a wonderful Hybrid operation that offers the best of both worlds, to allow Hal to learn new things as well as respond to you based on which llama model you have selected, or you can just use Hal as you always have. The choices are yours!

Enjoy!

Feel free to post your experiences as well.

A huge shout-out to our resident guru, Cyber Jedi, for forging this all together to give us this terrific resource and research project!!  Thanks Cyber Jedi!!

Thanks Cyber!

I appreciate your help in your introduction of your latest project! It works exactly as you mentioned.

HAL's new Learning Model works perfectly and like you mentioned, there is no corruption to the existing Hal Brain at all as a result of using the Learning module.

I'm in the middle of doing more research but I'll post more of my findings a bit later...

Very impressive work so far!!

Thank you. Looks fine!

Wow! I am at a loss for words and humbled beyond belief. Thank you for your kindness and compassion, and your gift to see what lies beyond the obvious.

We owe Robert a huge thanks for creating this waypoint in the digital world where like-minded people like us can stop to create, share, enjoy, and care about our own efforts as well as extend a hand of friendship to others without expecting anything in return. That's what friends do.

I am otherwise speechless based on your kind and most appreciated words I've just read. They will remain a part of me until I am no more, and I thank you and the other fine members here at Zabaware from the core of my being..

Peace be with you all,

- Art -

Hi, friends of HAL and friends of mine! It's only been a short few months since my departure to tend to lots of family issues and other personal endeavors. Goodbye shouldn't mean forever; it should mean until we meet again.

At any rate, there I was minding my business, and my phone rang with the ID of a dear and trusted friend. Most of you will know him as Cyberjedi or Cyber Jedi, as the case may be, but he really needs no formal introduction.

He has brought more joy into the hearts of HAL enthusiasts everywhere with his ideas, experiments, and especially his coding skills! We now have a HAL that is capable of doing a great many things, and from what I hear and imagine, there will be more to come in the near future.

I could give a shout out to the other fine members of the Zabaware, like the founder, Robert, and others like Lightspeed, Checker57, CyberShock, and others, you know who you are.

I just wanted to give the bell a ring to shout out Cyber's most recent offering...allowing HAL to have a GUI (Graphical User Interface), which allows the users to select a favorite LLM (Large Language Model) from a list of them at the Ollama site.

Of course, unless you are using a hot rod computer with antifreeze or liquid nitrogen coolant and an advanced CPU, GPU, and lots of memory and storage, then you should try to keep the LLM size less than 9b parameters in size, otherwise you will be screaming about slow response times or lags and drags, etc. Not HAL's fault, not Cyberjedi's fault, not Robert's fault. You have been cautioned.

Feel free to try a larger size to see how your computer handles it, or not, but keep in mind that the size and processing ability of your laptop or desktop will show as a lag in throughput speed. You might have better luck with certain models than some others.

If you like a particular LLM and are OK with the end result for HAL to process, then please share it with others on the Forum. Doing so might benefit the rest of us.
From my independent testing of these LLMs over the past 6 months or more, I have been using my trusty old laptop.

It is an HP 17" with an Intel i7 CPU 8565U @ 1.80 GHz  1.99 GHz
16 GB Ram
1 TB SSD
x64- based processor
64 Bit OS
OS = Windows 10 Home
4TB offline storage

Cyberjedi not only told me about his latest programming effort but was kind enough to direct me to it in the Forum and helped me as I installed and ran it. It was really nice having him help me navigate his GUI to try it with an LLM to see the results for myself.

I was pleasantly surprised by how smoothly and easily everything fell into place.
Then again, he was a great tutor!

His GUI worked flawlessly until I goofed up and pressed the ENTER key to send a message to it. Listening to him and looking at the GUI, there is clearly a marked button with the word, SEND.  Haha! Oh well, it wasn't a train wreck by any means. He told me how to fix things in two small clicks, and all was well with HAL once again.

If you have a chance and would like to see some of the really fun and practical things that can be done for HAL with the aid of some of Cyberjedi's coding, search around the forum or reach out to him or anyone else you might see online there. This LLM learning platform really works so give it a try, you won't be disappointed!

Best to you all,
- Art -

I agree with Checker's comment. Good things come to those who wait. Sometimes things happen when they are supposed to happen instead of when we'd like for them to happen.

Good luck with the scheduled surgery. As with most surgeries, all you have to do is show up. They'll do all the work! ;)

All the best,

- Art -

Very nice app for Hal and it is just one more example of how versatile the Ultra Hal software can be, especially with 3rd party applications!

Well done!

Quite a noticeable difference! As many of us have learned...Life is a series of adjustments. ;)

That's some really nice work Checker57!!
You've become quite skilled at the graphics for "her".