What it does
SafeCheckr makes legal documents easier to understand, to prevent human trafficking as promoted by contracts that traffickees do not fully comprehend or understand.
How we built it
Built on Flask, SafeCheckr utilizes natural language processing techniques and architectures to summarize text, and answer questions based on context.
Challenges we ran into
We originally used GPT-2 due to the fact that GPT-2 does not need much finetuning to work well. However, GPT-2 was giving more abstractive outputs for summarization and question-answering, so we had to pivot to using the transformers library and its array of architectures for these tasks.
Accomplishments that we're proud of
Our summarization and question-answering capabilities are certainly commendable. We started this project to make a difference, and we are really glas that we have the potential to do so.
Log in or sign up for Devpost to join the conversation.