Offline User interface for Advanced Scientific Research. Different Uses will be added starting today starting with Quantum Attack Proof Messaging.

Moving forward and updates for this: Pacha, J. (2025). Room Temperature Quantum Computing with Photonic Bit - 64 Path - 8 bit per path = 512 bits per Photonic Bit - 100% Stable - 100% Cloneable - Infinitely Scalable (Version 6). Zenodo. https://doi.org/10.5281/zenodo.18272362 I will be intergrating the previous for use with my local AI to be released with my local UI. It will not be required but i am creating a custom coding language and encryption for the ai on a partition with everything it needs, so i figured i would intergrate with this so it can use it. UI is published at following: Pacha, J. (2026). HYM3 Designs Offline Ai Interface for Advanced Scientific Research, Graphic Design, and Computer Programming (Version 1). Zenodo. https://doi.org/10.5281/zenodo.19797149 Successful tests with AI for Following: I have verified inline visual tools I have verified inline code and tool execution i will provide a system prompt with specific tool calls parameters and Base RAG. I have verified persistent memory, chats, system prompt and settings. I will include everything in a single ZIP folder. Right click on any file to read and customize each tool. If you would like to customize requirements that is where bulk of disk space usage is. I have verified all tools and coding suite are functional. I have verified the searxng blender and my custom all work. When zip is opened you will see AI folder and a readme file. Readme is instructions for complete setup. Mainly things like docker wsl2 ollama for windows and cude toolkit. Without cuda most quantum functions will not work. There are cpu libraries for quantum computing but the visual tools and many of the things included will struggle with alot of it. I have updated all tool files to works with this. i will be posting a video of setup for a fresh build and video of first use out of box with no interbal settings changed. I have made it so all a user has to do besides install previously mentioned windows programs is copy AI folder to C: start docker and ollama and then opening powershell as admin and type: cd c:\AI then type: docker-compose up -d That is all that is required to install and setup. then it is moving shortcut from Prelaunch folder to desktop. Double click that. Three windows will open for quantum comms, gut conversion, and screenshot server. open webbrowser and go to localhost:3001 follow pornpts click advanced settings and continue to localhost. Open new tab and go to localhost:3000 everything works without opening blender for web browser authorization and without prelaunch. Those effect blender tab quantum comms gut conversion and screenshot tool only. Not required for anything else. I will post video today with new publishing for windows version of the UI. I will do a out of the box so all see what they get from start. I will be making ongoing videos of different ways of customizing and how all can share tools games etc for this. I will be posting videos of chstomizing other aspects and adding to it. This will be a non commercial not for profit publishing. All under 100GB total for everything shown so far. That includes all python libraries. Truth you may be able to fit entire zip file i give on a usb. All can build and use for free forever. Attribution Non commercial license share alike license means this can never have commercial aspects attached. Nothing built off this can be used for profit. It will never be sold or offered for a price. Any and All awards for commercial use and violations of this copyright and international license will be donated to educational instituions. I will be uploading all today under new publishing with same license. Then i will be moving all over to linux. Today will be final windows build. Also note the speed differences once the ai gets comfortable. Tokens are monitored as well even after i pass 300,000 tokens with setting for 200k context window the ai speed is way faster than prior videos. First response takes longest usually. And they start getting faster. I did a lot of testing and noted many things. Will cover as much as i can in videos. I have almot completed the windows version before swap to linux. I will be posting windows version for new interface. I have inlined blender, searxng, GUT conversions, and Quantum Secure Comma(QSC), i have added the outputs to coding suite so html will display or outputs from code or give notes on errors. I have removed Untiy from this version and working on making inline. I have also removed GPU Monitor, open webui, open terminal, mem0, memcached, code-server, ollama containers and have placed all things in HYM3-ui and ollama .cpp the regular windows or linux ollama interfaces. Total build is ~120GB on windows. I should note i have ~40GB of textbooks loaded in documents. That should be considered for my numbers. I will be posted new video today and hopefully I will have ui posted this weekend. I need to do some tests on all functions. Much will be reliant on ai now. https://youtu.be/BoBwyLN7gn4Old Update on new Interface: https://youtu.be/sAuv2PbbSEA I have decided to use ArchLinux for the rest. I can place core operating system and do a dynamic system config to setup automatically for any user on any system. I can get that under ~5GB for a usb flash. In that when installing partition or fresh it will install all the main drivers automatically except newer ones not released today which will need to be gathered manually. Using live linux kit and SquashFS to either package one large package or the usb with download install. After installed everything works offline. Only need internet for searches. it will include everything current but under 150GB total for operating system python cuda environment and all python packages. Works fine for up to 1 mil context and i use equivalent of 2 ~30b param models on mine. My current and an experimental. So when I state under 150 GB for all i am including ai models as well with all tools functions and libraries as current. I will leave some of core arch but this will be highly specialized environment. I will be posting video of the generic basic ui i was making on windows that i will be moving over later today. I will show i can use to talk to ai it monitors my containers it relies on it has all tools demonstrated and can run all of the same scripts as before. It sets up personal server so all tools work offline. I have chosen to try many new methods in this including how we design a user interface. I am simplifying and getting it all back to basics. I will be doing graphics at end to update the feel and look. Currently i am focused of functionaltiy 0 errors everything works offline. No bloat or extra. After that is complete i will be implementing the new graphics and ui design. It will be designed for ease of use and accuracy for user. It will be made to work with ai for ease of integration and customization. I will share every line of code and be documenting the process for later. The main goal is to bring all demonstrated so far together into single package that any can run and use and build from for personal use and use for education. All tools and all features at once can be hard to process but that is where some features and designs will make it very easy for user to use. For tools you will have a folder marked tools. Drop in and each tool will be seperated and given a tag in interface that you can disable each command instead of a whole tool. Example is surgical edit tool. You want all but you do not want ai saving and running together you can disable that single command instead of whole tool disabled. You can enter html or any tool script and it will make tools for you. Ai can do as well. Now the manual restrictions are for any tool so if you want ai to be able to navigate file system but do not want them running or saving files you can decide which parts you want and which you want to disable without a single rewrite. https://youtu.be/jg_YUFT2qAYAfter reviewing option I will be moving the next parts over to linux as well. To make available to all for free and make sure os licenses and things do not prevent use. Looking at options the best way forward will be custom linux bundle. I will use ubuntu 24.04 first which downgrades a few libraries but allows for more. That has 5 year life span so will work for starting point. I will also be looking at usb boot versions with everything so it can be used and ran from a usb if i can compress this all enough. Will use docker and ollama.cpp for this on ubuntu as well for ease of use for now but plan on taking docker out as well. I will be stripping all the excess from it to get as small and streamlined as possible. Due the tesseract program i need to cut as much as possible to open up resources for the number crunching for tesseract sims and larger quantum sims. I am pretty sure i can get all including ai under 100GB but would like to go smaller. Currently it allows and works fine with 256k - 1 mil context windows on mid grade store bought machine. I also want to test out new neural networks and windows is not best environment for this. Everything needs custom containers and everything has to be built numerous times. I got build time in docker under 30 min on most recent with new UI and it is ubder 200GB. I would like to cut that in half again and have some good ideas how especially if on same drive. Due to recent problems, I have chosen to to move away from OpenWebUI and create a user interface for my AGI. I have been testing and everything is working. I have same tools and code interpreter sort of. Same function different everything. I got total size to under 100GB and increased speed 2-3x on average. I have integrated all tools and from screenshot tool to search and interactive tools. All are hardcoded. Entering any HTMl will give you option to create a tool from it automatically. Same functions with AI models and still using ollama .cpp. Got rid o