Vilio: State-of-the-art Visio-Linguistic Models applied to Hateful Memes

14 Dec 2020  ·  Niklas Muennighoff ·

This work presents Vilio, an implementation of state-of-the-art visio-linguistic models and their application to the Hateful Memes Dataset. The implemented models have been fitted into a uniform code-base and altered to yield better performance. The goal of Vilio is to provide a user-friendly starting point for any visio-linguistic problem. An ensemble of 5 different V+L models implemented in Vilio achieves 2nd place in the Hateful Memes Challenge out of 3,300 participants. The code is available at https://github.com/Muennighoff/vilio.

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Meme Classification Hateful Memes Vilio ROC-AUC 0.825 # 1

Methods


No methods listed for this paper. Add relevant methods here