AIC CTU@FEVER 8: On-premise fact checking through long context RAG

TL;DR

This paper introduces an on-premise fact-checking pipeline based on a two-step RAG approach, achieving top performance in FEVER 8 with limited computational resources and time constraints.

Contribution

It demonstrates how to deploy a state-of-the-art fact-checking system on limited hardware, improving practicality and accessibility.

Findings

Achieved first place in FEVER 8 shared task.

Maintained high performance with only one NVidia A10 GPU.

Operated within 60 seconds per claim.

Abstract

In this paper, we present our fact-checking pipeline which has scored first in FEVER 8 shared task. Our fact-checking system is a simple two-step RAG pipeline based on our last year's submission. We show how the pipeline can be redeployed on-premise, achieving state-of-the-art fact-checking performance (in sense of Ev2R test-score), even under the constraint of a single NVidia A10 GPU, 23GB of graphical memory and 60s running time per claim.