bb.donnay-software.com

Hi All

Will xBase++/Express++ be suitable for Artificial Intelligence projects now and in the future ?

Thanks

Joe

hi,

unixkd wrote: ↑Tue Jan 30, 2024 1:56 pm Will xBase++/Express++ be suitable for Artificial Intelligence projects now and in the future ?

as long as Xbase++ use 32 Bit App it is not possible to use "big" AI-Model > 4 GB loading into RAM

We're using an AI service - provided by a partner company - in our mobile app. The service is mainly used by our backend (Xbase++, Xb2.Net) which connects to it and sends audio data from the app. The audio data is analyzed and turned into structured medical information. No magic, works excellent (> 95 percent). Maybe it would be hard to integrate such kind of service directly into our backend, but there's no need for that kind of approach. We are currently designing more use cases for that kind of technology.

A) Native: No. It is simply not adequate for AI development as it lacks just about every needed tool to do it in code.
Python has great and easy to use libraries. Java too.

B) Connect to existing services via REST/SOAP/socket : YES and it can be used already (see @Tom post).

So if you write pyton code and put it onto application server , you can definetly call it via one of protocols. Might even develop own or integrate existing tools into it.

unixkd wrote: ↑Tue Jan 30, 2024 1:56 pm Hi All

Will xBase++/Express++ be suitable for Artificial Intelligence projects now and in the future ?

Thanks

Joe

Hi,

It's available right now: https://platform.openai.com/docs/api-reference.

Hi all,

@Jimmy: The purpose of a forum is to get help. This means that users are responding to questions with SOLUTIONS! If you don't have an answer to the question, then don't answer. Thanks in advance.

hi Chris,

i´m talking about "local" AI-Model "trained" with own Data which need much RAM

i do have a Fivewin and HMG native Solution but it must be 64 Bit App to use llama64.dll

you can get Source using "Github Desktop" and type
"git clone https://github.com/ggerganov/llama.cpp"
and make some change (see Link to Fivewin Thread) to compile as *.DLL

the "Problem" is that the DLL must build on that PC which you want to use with your own "Local" AI-Model
https://forums.fivetechsupport.com/view ... hp?t=44028

---

Antonio also descript how to use own Data to "trained" own AI-Modal e.g. all News of Forum
but it need huge Power which strong PC and GFK like RTX 4xxx using Python

Hi Jimmy,
that is not relevant for the question:

Will xBase++/Express++ be suitable for Artificial Intelligence projects now and in the future ?

hi Wolfgang,

Will xBase++/Express++ be suitable for Artificial Intelligence projects now and in the future ?

i think only "local" own AI-Model using own Data will work correct

you can compile Source of llama.cpp for 32 Bit DLL but that does not make much Sense
it will general work but "local" AI-Model need much more RAM than Xbase++ can give you using 32 Bit

Hi,
An example of AI project developed in Xbase++ is the AIDOS system created by professor Eugene Lutsenko. Its purpose is Cognitive Analytics/Analysis, which applies human like intelligence to certain tasks. In its work, Cognitive Analytics combines Artificial Intelligence and Data Analytics.

By definition, Cognitive Computing is: “A subfield of artificial intelligence which simulates human thought processes in machines using self-learning algorithms through data mining, pattern recognition, and Natural Language Processing”. The AIDOS system of professor Lutsenko works that for many different tasks. It is more described in the following post, and it is free to download and use it from professor's site.

viewtopic.php?f=7&t=1443

As you see, it is possible to develop AI system in native Xbase++ no matter is it a 32/64 bits application or how many RAM do you have.