Core Concepts
Team Trifecta's Pre-CoFactv3 framework excels in fact verification, surpassing competitors and setting new standards.
Abstract
Team Trifecta's Pre-CoFactv3 framework, focusing on Question Answering and Text Classification components, achieved remarkable success in the AAAI-24 Factify 3.0 Workshop. By leveraging In-Context Learning, Fine-tuned Large Language Models (LLMs), and the FakeNet model, the team secured first place by surpassing the baseline accuracy by 103% and maintaining a 70% lead over the second competitor. The research paper explores various approaches, including comparing different Pre-trained LLMs, introducing FakeNet, and implementing ensemble methods to enhance fact verification accuracy. The success of Team Trifecta underscores the efficacy of their approach in advancing fact verification research.
Stats
Our team secured first place in the AAAI-24 Factify 3.0 Workshop.
Surpassed baseline accuracy by 103%.
Maintained a 70% lead over the second competitor.