Leverage AI to revolutionize education.
Learn how AIs work while creating ur own local ai system based on RAG
Even on limited educational material (private information)
SCROLL DOWN!
Large Language Models (LLMs) are AI models trained on vast amounts of internet data. For example, ChatGPT is an LLM. But what if you could run such a model entirely offline, on your own machine, without internet access?
Well, you can—thanks to open-source LLMs!.
Retrieval-Augmented Generation (RAG) enhances AI's ability to retrieve and generate responses based on specific data you provide. Imagine you have a large dataset—such as a 500-page PDF—containing valuable information. Instead of relying on internet searches that may yield generic or inaccurate results, RAG ensures the AI retrieves and utilizes only the relevant information from your provided dataset. This approach improves accuracy, relevance, and contextual understanding, making AI responses more reliable and domain-specific.
Implementing RAG involves integrating a retrieval system that fetches real-time data from the datasource before the AI responds with set parameters that you give it.
This could be , the complexity or the way u want ur Ai to think or the way you want it to give output to you
My first model M1000 is a simple RAG system which uses Tinyllama as its LLM ,Unlike default rag system i have made chunks of the input data before storing it into a vector database with the chunks overlapping one another, this is my definition of a basic and efficient RAG system.
Fine-tune retrieval methods, embeddings, and ranking strategies to enhance efficiency later on when u continue building ur retrieval pipeline. M2000 my next vertion of the bot will be focused on ranking and indexing the data more efficiently.
Venturing into the world of AI is not a simple task expecially if u dont find the right community my goal is to help styudents new to rag and its systems get to know eachother and their projects better this helps us bing a sence of community and overall inspire people to do more and to be inspired by others projects and grow together!.
Making ur own Localized RAG system comes with its own benifits and i think you should give it a try too!
My first AI will be named M1000 and its first version will be 1.0.0
I plan on giving him a personality and focus on its prompting later on
:) check out the pipeline rag diagram for the M1000 below , Github linked down below! read the READ ME file to get an entire explanation on how M1000 works (ps its highly detailed its perfect if ur a beginner)
View on GitHub