YouQuiz

Inspiration

My inspiration to create this app in school during the breaktime. In school we don’t have access to internet so I wanted to creat a app that uses local LLMs to generate quizes with RAG system. I wanted to watch the videos that I download at home and than take quizes on specific parts of that lecture video. That is how I came up with the idea of YouQuiz and now my whole class uses it. Even my teachers!

What it does

First it transcribes voice of youtube video into a text. Then that text is seperated into chunks and then embedded so that the LLM can process the text. Afterwards from ollama you start a local server than may enter your prompts which is than processed by the LLM of your choice. In the end the LLM generates you a quiz just from a Youtube URL. An example youtube video is included in the repository as well.

How we built it

I used Retrieval Augmented Generation (RAG) Ai technique which is improving day by day. As you can see from my other repositories on Github I am an enthuasiast about artificial intelligence and love implementing it to daily life. I also have experience in data analytics such as gathering necessary metrics and analyzing data. With this knowledge I tested different optimizations (temperature, top_k, top_p) and determined that the best output is via the current settings. I noted down the token numbers the LLM generates with different prompts and also determined that this is the best prompt for most relevant answers. If your PC is strong you may use deepseek-r1:8b model as well whichs trained on 8 billion parameters. Finally I added necessary notes on the code as well, explaining the processed and blocks. Also of course I could use othee embedding models or databases like ChromaDB but these are the best setting that help me acquire the best results and I have limited computer RAM. If you have an even weaker PC than mine I suggest Tinyllama model of Meta or if you have a strong PC you may try local Gemma models by google. Enjoy YouQuiz!

Challenges we ran into

My main challenge was writing the base prompt for this application. I signed in for a prompt-engineering course by Vanderblit University and used Question refinement patterns to achieve the best prompt. I also decided to implement the persona pattern and chain-of-thought prompting into my prompt as well to get the best results. That’s how I managed to create a well balanced prompt which gives relevant outputs!

Accomplishments that we're proud of

I am proud that although I have a GPT-4o mini API key I worked hard to create a LLM AI system which is free to use for everyone and is also easy to use. I am proud that I have discovered new tactics in RAG industry and shared it with every programmer like me by having it as an open source project. I am grateful to HackForge team for giving me this opportunity!

What we learned

I learned more enhanced coding of RAG AI systems which is handy in analyzing huge texts, pdfs and even Youtube videos in my case. I also gained new precious knowledge about prompt-engineering and how we should interact with Large Language Models. With this project I was introduced to the huge world of Langchain as well. I hope I can further improve my skills and contribute to great projects before applying for college. Also I gained deeper knowledge about Hackhathons and how they can play a crucial role in university admissions. Thanks again to HackForge team for providing me this chance to prove myself.

What's next for YouQuiz

After I reciece feedback from sector professionals with the help of the HackForge hackhathon, my main aim is to improve the UI of YouQuiz. If I manage to learn more about programming UI and improving UX, I believe the future of YouQuiz is really bright. Also I need to get a new computer to test LLM models and langchain modules with higher RAM requirements :) Furthermore I am planning to implent vision-based AI systems to analyze youtube videos with their visuals as well. If I work hard and overcome these obstacles I want to launch YouQuiz as an offline mobile application! Waiting for your judgement!

Built With

• ollama
• python