HackStack

Project Info

Quizzical

Devpost

This project did not submit a demo video on Devpost.

Inspiration

The inspiration for our transformative study tool emerges from the universal challenges students encounter when sifting through extensive information to identify the most relevant study material. As students ourselves, we often grapple with the task of making our own notes for each and every class while trying to manage the inherent stress of dorm and college life. Traditional methods can sometimes miss the mark, either bypassing crucial concepts or because of the use of presenting materials that don't align with individual comprehension levels. We aim to devise a system that goes beyond merely delivering information, but rather engage, respond, and progress with the student. While the adaptive learning realm in educational technology provides some cues, our game-changer was the integration of generative AI to convert lecture notes or videos students provide into tailored, interactive study aids. The goal isn't merely to assist with information but to shape the material based on a student's unique learning curve, ensuring every topic or concept fits like a puzzle piece in their cognitive map. Our approach isn't just about efficiency; it's about redefining the study experience, harnessing technology's power to cater to each student's needs - a study revolution for students, by students.

What it does

Building on this vision, the heart of our system lies in its ability to transform traditional lecture notes or videos provided by students into dynamic, personalized study aids through generative AI. Rather than wading through a static set of notes, students experience a responsive learning environment where content evolves to meet their unique needs. Quizzical allows the user to upload any study material in either video format or pdf format. Our application synthesizes all of the content in the study material allowing for the creation of flashcards and concise summaries. Our approach streamlines the learning process, ensuring that students don't just study, but they engage with material fashioned specifically for them, offering a blend of efficiency and individualized learning.

How we built it

MERN Stack (MongoDB, Express.js, React.js, Node.js) Together.ai API HuggingFace Model llama-2-70b-chat Whisper.ai

Challenges we ran into

Integrating multiple technologies like the MERN Stack, Together.ai API, HuggingFace, and Whisper.ai, while ensuring they all communicated flawlessly, proved to be complex. Each tool had its nuances and required specific configurations to function harmoniously within our system. The translation of lecture videos into structured data via Whisper.ai posed particular difficulties, especially with the extended processing time required to accurately convert the content. Furthermore, optimizing the HuggingFace model, llama-2-70b-chat, to ensure the generated study material was both relevant and accurate, demanded continuous fine-tuning. Despite these obstacles, our commitment to delivering a transformative study tool drove us to innovate, iterate, and eventually overcome each hurdle.

Accomplishments we're proud of

One of our most notable achievements is the system's versatility in handling various formats of learning content. We successfully engineered our tool to accept both PDF lecture notes and videos, ensuring that students have the flexibility to upload their preferred study materials. This adaptability not only broadens our user base but also caters to diverse learning preferences. The transformation of these formats into tailored study aids showcases our dedication to creating a truly comprehensive and inclusive study platform.

What we learned

We learned a lot about LLMs and Gen AI through this project.

What's next

In our future developments for this project, we aim to further harness the power of video, elevating the potential of our AI to generate diverse study materials tailored to the three primary learning styles: visual, auditory, and kinesthetic. Recognizing that each student has a unique learning preference, the system will be designed to interpret the provided content, whether it's textual lecture notes or videos, and transform them into a format that resonates best with the user's style of learning. Furthermore, the advanced use of facial expression analytics will play a pivotal role; as students interact with each flashcard, the AI will detect signs of confusion. Based on these nuanced emotional cues, the system will respond in real-time, fine-tuning and generating content that addresses the specific areas of challenge, ensuring a comprehensive and adaptive learning experience for all.

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • CSSIn code
  • ExpressIn code
  • HTMLIn code
  • JavaScriptIn code
  • LangChainIn code
  • MongoDBIn code
  • PythonIn code
  • ReactIn code
  • Hugging FaceClaimed
  • Node.jsClaimed

8 of 10 appear in the indexed code. 2 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

38 KB

Source files

21

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars