General Track and Fastly Track
Inspiration
In the contemporary world, we must visit several websites to access several research papers for our needs. Then comes the issue of understanding and parsing through the paper. For that reason, we devised a solution to create an all-inclusive website incorporating all these features.
What it does
The website we built allows users to find research papers using our search features. You can filter out searches based on relevance, DOI, etc. Furthermore, you can summarize the paper and get a similarity percentage for the report we decided on.
How we built it
We built a website with a backend of Flask. The front end was kept simple and built using HTML, CSS, and Javascript. We used the arxiv API for fetching the ChatGPT features to summarize.
Challenges we ran into
The length of the paper exceeds max input size of the model. Resolved by creating batches of paper. Cross-browser compatibility issues Struggled with frontend Teammates fell asleep
Accomplishments that we're proud of
The ChatGPT summary worked better than we expected. We could learn more about the API and this being our first international hackathon, we were proud that we could complete the scope of the project.
What we learned
We learned how to use open-source APIs to the full extent.
What's next for SmrzGPT
Include text-to-speech features. Include translation for the summary. Make it open source. Monetize it.
Log in or sign up for Devpost to join the conversation.