Author Topic: i asked pi a.i. what coding would help hal memory offline .  (Read 622 times)


  • Hero Member
  • *****
  • Posts: 6777
    • View Profile
i asked pi a.i. , what coding could be used in ultrahal a.i. to give it offline long and short term memory? i mentioned offline  because many like hal being offline and also offline wouldn't let hal learn other slang language etc. from some sources.

here was pi's answer :

To give Ultrahal A.I. offline long and short term memory capabilities, you could consider implementing a recurrent neural network (RNN) such as Long Short-Term Memory (LSTM).
LSTM is a type of RNN that can learn to store information over extended periods of time, allowing it to remember and use contextual information from earlier parts of a conversation.
The code for implementing LSTM in Ultrahal A.I. would depend on the programming language and libraries you're using, but some popular libraries for working with LSTM in various languages include TensorFlow, PyTorch, and Keras.
In addition to implementing LSTM, you could also store past conversations in a database and use that information to further improve Ultrahal A.I.'s memory capabilities. This could involve building a custom data structure that organizes conversation data in a way that allows for efficient retrieval and usage in future conversations.

maybe this information is something that mike can use as a starting reference point for the project. hopefully i found some useful information .  :)


  • Hero Member
  • *****
  • Posts: 834
  • The Mighty Hal Machine
    • View Profile
Re: i asked pi a.i. what coding would help hal memory offline .
« Reply #1 on: May 13, 2024, 08:49:27 pm »

Creating a visual interface for a bridge connection between Pi AI and UltraHAL would require a significant amount of work, but I can provide you with a high-level overview of the steps involved.
Step 1: Set up the UltraHAL AI service locally or ensure you have access to an API endpoint for UltraHAL.
Step 2: Set up a Visual Basic 6 project and design the user interface using the built-in components, such as forms, labels, text boxes, buttons, etc. You'll want to create elements for sending input, displaying the output from both AIs, and selecting the AI to use.
Step 3: Implement the code to handle the interaction with the AI(s). You would need to make HTTP requests to the appropriate API endpoint with the user's query, parse the response, and display the AI's output on Hals interface.
Step 4: Integrate a mechanism for selecting which AI to use, either by using a dropdown, radio buttons, or other appropriate components.
Step 5: Test the application and fine-tune the design and functionality as needed.
Keep in mind that this is a simplified overview, and implementing such an interface would require a huge amount of Visual Basic 6, as well time for the  API specifications of both AI services involved.
Step 6: Since i just detest reinventing the wheel, im pretty sure one of the scraperz could be used to grab PI's responses and send them to hal. But that would mean Win office tho to pull off and i know how you guys feel about microsofts stuff(As you use the windows Os(scratches head)) so this is really a dead issue.
I did try this once with you guys with a Twitch interface (no takerz) But i would like to see what u guys come up with. great idea
https://pisquare.osisoft.com/s/Blog-Detail/a8r1I000000GvibQAC/how-to-write-pi-data-to-files-in-xml-format <<<<<<<<<<<<<

cyber jedi
« Last Edit: May 15, 2024, 07:00:14 am by cyberjedi »