Introduction
In the vibrant world of hackathons, where innovation knows no bounds, our team, OSSquad, had the privilege of participating in a national hackathon hosted by the Disha Committee at COER University.
Our mission? To develop a cutting-edge AI platform that offers a crucial service: text summarization. We proudly present to you, InfoGPT.
The Technologies Used
Our journey to create InfoGPT began with a critical choice of technology. To harness the power of text summarization, we opted to use the BART-CNN model released by Facebook. This sophisticated model, which combines BERT and CNN, showed great promise for our project.
But using the model was not enough; we needed a reliable way to call it into our platform seamlessly. This is where the Hugging Face Inference API came to our rescue. It allowed us to access and utilize the BART-CNN model efficiently.
In our quest to make InfoGPT user-friendly, we integrated Postman, a popular API testing tool. Postman made it easy for users to interact with our platform, providing a straightforward interface to input text and generate summaries.
User Experience
InfoGPT was designed with the user experience in mind. Users could input text ranging from as short as 200 characters to as long as 100,000 characters. Whether you needed a concise summary of a tweet or an extensive overview of a research paper, InfoGPT had you covered.
The generated summaries were tailored to the user’s requirements, with lengths varying from 300 to 1,000 words. This flexibility ensured that InfoGPT could serve a wide range of users, from students to professionals and beyond.
We took user experience to the next level by allowing users to choose between different models for text summarization. While the BART-CNN model was the foundation, we also integrated Google’s Flan-T5-XXL model to offer enhanced summarization options. Choice is power, and we wanted our users to have it.
Feature Extensions
Our original plan for InfoGPT included document summarization using the arc53 model and text-to-image capabilities through the stable-diffusion. However, as is often the case in hackathons, time constraints became a challenging adversary. We were unable to implement these features during the competition, but the potential for future development is undeniable.
Incorporating Google’s Flan-T5-XXL Model
We decided to add Google’s Flan-T5-XXL model to InfoGPT, further enriching our text summarization capabilities. This addition offered our users even better results, highlighting our commitment to providing high-quality AI-driven text summarization.
Code Snippets
Let’s dive into the technical aspects of InfoGPT. Here are some code snippets to help you understand how we called the BART-CNN model and allowed users to select their preferred summarization model:
Calling the BART-CNN Model from Hugging Face API
let config = {
method: 'post',
maxBodyLength: Infinity,
url: 'https://api-inference.huggingface.co/models/facebook/bart-large-cnn',
headers: {
'Content-Type': 'application/json',
'Authorization': 'Bearer ' + process.env['Access Token']
},
data : data
};Conclusion
InfoGPT has been a thrilling project that highlights the potential of AI text summarization. It serves as a testament to the power of innovation, even in the face of time constraints. As we reflect on our journey, we are excited about the future of InfoGPT and the possibilities in the field of AI text summarization.
We would like to express our gratitude to our team members for their dedication and hard work, COER University for hosting the hackathon, and the Disha Committee for organizing this wonderful event.
We invite you to experience InfoGPT for yourself. Try out our platform, generate text summaries, and provide us with valuable feedback. Your input will help us further refine and enhance InfoGPT.
[Link to InfoGPT Platform — Coming Soon]
Final Thoughts
In the world of hackathons, where time is limited and challenges are boundless, InfoGPT emerged as a testament to the innovation and the power of teamwork. We are excited to see how our platform evolves and grows, and we invite you to be a part of our journey.
Thank you for joining us on this remarkable adventure. We look forward to your support and feedback as we continue to push the boundaries of AI text summarization with InfoGPT
