Question 1

What are Semantic Textual Similarity datasets used for?

Accepted Answer

Semantic Textual Similarity datasets are collections of labelled or raw data used to train, fine-tune, and evaluate models on the semantic textual similarity task. This page lists 5 such datasets, each linking to its source and paper.

Question 2

Which Semantic Textual Similarity dataset is best for benchmarking?

Accepted Answer

None of the listed Semantic Textual Similarity datasets are currently tracked as standard LLM benchmarks, but many are widely used for evaluation.

Question 3

How many Semantic Textual Similarity datasets are there?

Accepted Answer

We catalog 5 Semantic Textual Similarity datasets in one searchable directory.

Semantic Textual Similarity Datasets

What languages do semantic textual similarity datasets cover?

Explore other dataset tasks

Frequently asked questions