Skip to content

Dataset · sentence-transformers

Combined NLI corpus for embedding training

sentence-transformers/all-nli

Millions of English sentence pairs with logical inference labels for training semantic similarity models.

What's inside

Contains English text pairs—premise and hypothesis—combined from known SNLI and MultiNLI datasets. Each pair is assigned one of three classes: entailment, neutral, or contradiction. Corpus size ranges from 1 to 10 million examples.

What you can do with it

  • Training and fine-tuning embedding models to assess semantic text similarity
  • Training classifiers for natural language inference (NLI) tasks
  • Using as a base corpus for duplicate search and reranking in RAG
Tasks
Embeddings, Sentence similarity
Languages
English
Size
1M–10M rows
Downloads this month
4.5K
Likes
54

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("sentence-transformers/all-nli")