3 datasets found

The ReCoRD dataset is a benchmark for reading comprehension tasks.

The dataset used in the paper is the SNLI dataset.

Pre-trained language models (PrLM) have to carefully manage input units when training on a very large text with a vocabulary consisting of millions of words. Previous works have...