Primer

Primer Logo

We present “Primer” as an educational platform that brings accessible high-quality education across linguistic and socioeconomic barriers.

Our Mission: To bring accessible education to everyone.

Inspiration

Internet and Schooling

Access to the internet is positively correlated with years of schooling.

But it’s not the internet itself improving education, it's the easy access to information that allows for academic growth. Providing that information in a format accessible to students of all backgrounds, with or without information, will start bridging that gap and aiding our global education crisis.

What it does

Primer is an educational platform powered by generative AI designed reduce educational inequalities accessible on low-end consumer hardware. Leveraging various generative machine learning models, we are able to generate high-quality content in native languages and accessible formats that is both contextually relevant and tailored to the curriculum needs.

Key Features:

Customizable Learning: The platform can adapt content based on specific student needs and educational goals.
Multimedia Support: Generates and delivers content in various formats, including text, audio, and images, to cater to different learning styles.
Scalability: Designed to work with both cloud-based and local compute resources, ensuring scalability and flexibility.
Accessibility: Optimized to function on lower-end hardware, making it accessible to schools with limited technological infrastructure.

Who can use it

This friendly platform is suitable for anyone, including:

Educators seeking to supplement teaching with AI-generated content.
Students who require additional educational support outside of the classroom.
Knowledge Seekers looking for a easy way to explore new information for fun.

What's next for Primer?

The next steps for Primer are:

Implementing the visual designs illustrated in our mockups
Reducing the size of the program to make it even more accessible
UX Testing with a variety of audiences

United Nations' Sustainable Development Goals

Primer addresses the following United Nations' Sustainable Development Goals:

GOAL 4:

Ensure inclusive and equitable quality education and promote lifelong learning opportunities for all

GOAL 5:

Achieve gender equality and empower all women and girls

GOAL 8:

Promote sustained, inclusive and sustainable economic growth, full and productive employment and decent work for all

GOAL 10:

Reduce inequality within and among countries

GOAL 16:

Promote peaceful and inclusive societies for sustainable development, provide access to justice for all and build effective, accountable and inclusive institutions at all levels

GOAL 17:

Strengthen the means of implementation and revitalize the Global Partnership for Sustainable Development

Our Team!

Our Team

Primer Bot

How we built it

Backend

Hugging Transformers, Accelerate, Optimum, Flash Attention, and Quart

Front End

React Native Expo, Tailwind

Data + Training Pipeline

Data Collection: We scrape open-source educational content from sources like OpenStax to create a diverse dataset.
Dataset Creation: A custom dataset is generated focusing on instruction-following capabilities using few-shot learning techniques with LoRA.
Model Fine-tuning: LLMs are fine-tuned with the custom dataset to enhance their understanding and generation of educational content.
Validation: Rigorous testing and validation of the LLMs ensure the accuracy and quality of the generated content.

Inference Pipeline

Input Parsing: Student inputs are parsed and interpreted using a combination of speech-to-text services (Whisper) and automated parsing tools.
Prompt Generation: AI generates prompts based on the parsed input to guide the LLM towards generating the required educational content.
Content Generation: The LLM uses retrieval-augmented generation (RAG) techniques to create detailed, accurate educational content.
Scoring and Resampling: The content is scored and resampled for quality and relevance.
Output Parsing: The generated content is parsed into different formats (text, audio, images) to match the preferred learning mode.
Delivery: The content is delivered back to the student, with options for text-to-speech conversion or other multimedia formats.