The viewer is disabled because this dataset repo requires arbitrary Python code execution. Please consider removing the loading script and relying on automated data support (you can use convert_to_parquet from the datasets library). If this is not possible, please open a discussion for direct help.

This dataset is a quick-and-dirty benchmark for predicting ratings across different domains and on different rating scales based on text. It pulls in a bunch of rating datasets, takes at most 1000 instances from each and combines them into a big dataset.

Requires the kaggle library to be installed, and kaggle API keys passed through environment variables or in ~/.kaggle/kaggle.json. See the Kaggle docs.

Downloads last month: 45

Edit dataset card