Open Source, High Function AI Assistants Running Locally

NOTE; THIS IS AN ALPHA BUILD WITH THE FOCUS OF TESTING THE TECHNOLOGY BEHIND THE PROGRAM, ONCE CONFIRMED I WILL MOVE ONTO ADDING MORE FUNCTIONALITY AND LOOKS! ALSO PLEASE LOOK AT THE INSTALL HERMES SECTION RIGHT BELOW THIS, ELSE THE PROGRAM WON'T WORK

Project Replicant is an attempt to bring high functionality AI to users completely locally. As long as you have a decent computer that's Windows capable, you can run ChatGPT level LLMs without any fees, licenses or even an internet connection!

Project Replicant utilizes the Arhua-Sys (Or the Artificial Reality Human Understanding and Assistance System); a subset of systems meant to bring LLMs to the next level.

NEW UPDATE COMING,  WINTER 2024

Please Install Hermes to Use This!

Please Install Hermes from (https://huggingface.co/TheBloke/OpenHermes-2.5-Mistral-7B-GGUF/blob/main/openher...) and drop it at (Project Replicant 1.1a\Project Replicant_Data\StreamingAssets)! Huggingface will download the file quicker than Itch.io takes to download the rest, without this singular file the program will not work!

Completely Open Source

Project Replicant is built atop open source technologies and is completely free to use for all (Except Meta, tech mongols or companies affiliated with them, who is banned from using any of these tools as well as any hate groups)!

Project Replicant utilizes technologies such as;

LLMUnity (A package for using LLMs in Unity)              https://github.com/undreamai/LLMUnity

and

Hermes 2.3                                                                        https://huggingface.co/teknium/OpenHermes-2.5-Mistral-7B

After Alpha, the code for Project Replicant will be dropped on the Walker Industries Github;

https://github.com/orgs/Walker-Industries-RnD/repositories

Long Term and Session Memory

This LLM uses a summarizer as well as a custom AI/English format called Engels to save information! The summarizer takes the key notes from conversation while Engels shortens it within a non grammatical format understandable by ANY capable LLM!

The system within also compresses information over time and saves it between sessions!

Completely Uncensored

Gone are the days of "appropriate" filters! Want to debate on the meaningless of the current day as humanity slowly moves towards an irreversibly screwed future due to the indifference those in power hold? Or maybe you want to talk about something hotter...

In either case, the AI is completely uncensored and will listen to whatever you have to say and give you it's best response!

Future Plans and Milestones

As you can probably tell, there isn't too much here that makes this AI special (After the fact that it runs near GPT4 level power while being completely local)! Some changes in future versions (Some of which have already been partially implemented) will include;

1. The usage of WhisperX for automatic voice and speech recognition
2. The usage of RVC for AI voices
3. Changing the summarization/Engels system from being LLM based to completely NLP based
4. Integrating video and images (Eventually video at a realtime speed)
5. Creating a 3D, VR mode for the AIs to act similarly to Johnny Silverhand from Cyberpunk 2077
6. A note system for remembering information said by the user
7. Encryption with AES256 and Scrypt
8. Importing custom AI personalities
9. Using your AI with other humans and AIs and 
10. Being able to play games with your AI

It's a long road but half of these are partially implemented already!