Inspiration

History classes can sometimes feel tedious, yet the subject itself holds fascinating stories and valuable insights when visualized effectively. With the right approach, history transforms from dates and facts into vivid narratives that capture imagination, making learning engaging and revealing the richness of past events and cultures.

What it does

His.Story is an innovative app that transforms educational content by taking any user query, retrieving relevant information directly from PDFs, and explaining it through dynamic animations. Powered by advanced language models like GPT-4, the app refines the retrieved text, converting it into a structured JSON format that integrates seamlessly with the app’s functionality. GPT-4 also enhances the narrative, breaking down complex sentences into clear, student-friendly explanations, fostering easier comprehension. Through this process, His.Story combines text processing, animation, and engaging storytelling to deliver a highly interactive learning experience that makes challenging subjects accessible and enjoyable for students.

How we built it

His.Story is built using advanced large language models, including GPT-4 and LLaMA. First, a query is processed through a LangChain-powered Retrieval-Augmented Generation (RAG) system to extract relevant content. This text is then refined by GPT-4, while LLaMA generates prompts for image creation later to the image generation model. The resulting visuals and text are combined with speech synthesis to produce a cohesive, animated lecture. This multi-model pipeline creates an immersive, interactive learning experience, enhancing students’ understanding and engagement with history.

Challenges we ran into

The current 3-4 minute delay in His.Story, primarily due to image generation processing, is a noted area for improvement. With the integration of InferallAI, we anticipate significant reductions in this delay. Another challenge is ChatGPT’s tendency to extend explanations, potentially leading to overly lengthy content for concise queries. While prompt adjustments have mitigated this issue, there remains a chance of occasional verbosity. Continuous refinement is underway to ensure efficient, clear responses that maintain students’ engagement and optimize their learning experience.

Accomplishments that we're proud of

What we learned

What's next for His. Story

Built With

Share this project:

Updates