We only provide new dev and test sets for reproducing our result. The Train sets are not in this repository because of their big sizes, you can obtain them from the links below: Yelp Sentiment: You can get it from the [CAE Github](https://github.com/shentianxiao/language-style-transfer/tree/master/data/yelp) Yelp Gender: Follow instructions on the [Privacy-Aware Text Rewriting](https://github.com/xuqiongkai/PATR) repo or [here](https://github.com/dadelani/PATR). We are making use of the first 100,000 sentences for male and female gender as the train set.

The dataset contains Yelp reviews, each review has both gender and sentiment annotation. The texts are obtained from two datasets (YelpGender & Yelp Sentiment) but from the same source. By automatically comparing each review in the test set of YelpGender with the YelpSentiment Dev and Test sets, we created a new Dev set and Test set with 4,000 reviews, each with both gender and sentiment information. This can be used for future research to evaluate the utility of the Yelp Gender dataset.