4 datasets found

The dataset used in the paper is a semi-verified crowdsourced learning dataset, where the majority of the crowd workers are adversarial and the rest behave as Massart noise.

The Good Judgment Open (GJO) dataset contains 1770 datapoints (698 'forecasts' and 1072 'comments') posted by 242 anonymised users with a range of expertise.

Danbooru2017 is a large-scale crowd-sourced and tagged anime illustration dataset.

The dataset used in the paper is the Stanford Natural Language Inference (SNLI) dataset, which consists of 549,367 premise-hypothesis pairs for train/dev/test sets and target...