AI Tools.

Search
feature extraction by intfloat

e5-mistral-7b-instruct

E5-Mistral-7B-Instruct is an embedding model that leverages the full generative capacity of Mistral 7B by using decoder-only LLM representations for text embeddings. It uses instruction prompts at inference time to orient embeddings for retrieval, clustering, or classification tasks. At release it achieved state-of-the-art MTEB scores for dense retrieval, outperforming BERT-family embedding models by a significant margin on hard retrieval tasks.

Summary text generated by an automated pipeline from the model card · Not individually reviewed or run by us · How this page is made

Share: X Reddit HN LinkedIn

From the model card

Fields below are copied from the tags and counters on the HuggingFace repository intfloat/e5-mistral-7b-instruct at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.

Publisher (HF namespace)
intfloat
Pipeline tag
feature-extraction
Library
Sentence Transformers, Transformers
Framework tags
PyTorch
Weight formats
safetensors
License tag
mit — read the license file in the repo before relying on it
Language tags
English (en)
Papers cited
arXiv:2401.00368, arXiv:2104.08663, arXiv:2210.07316, arXiv:2212.03533
Downloads (HF counter at last fetch)
520,252
Likes (HF counter at last fetch)
569
Model card
https://huggingface.co/intfloat/e5-mistral-7b-instruct

Use cases

  • High-quality dense passage retrieval for RAG systems requiring top embedding performance
  • Long-document semantic similarity where BERT-based models truncate
  • Asymmetric retrieval tasks (short query, long passage) using task instructions
  • Academic benchmarking of LLM-based vs encoder-based embeddings
  • Building search systems where retrieval quality justifies 7B inference cost

Pros

  • MTEB-leading performance for dense retrieval at time of release
  • Instruction-steered embeddings adapt to retrieval, classification, or clustering tasks
  • 564 likes with broad adoption in high-quality RAG applications
  • Apache 2.0 license; text-embeddings-inference compatible

Cons

  • 7B parameters require a GPU; 10-100x more compute than BERT-family embedding models
  • Inference latency is high; not suitable for real-time embedding of short texts at scale
  • MTEB rankings evolve quickly; newer models (NV-Embed, GTE-Qwen) now score higher
  • Requires specific instruction prefixes; generic usage without instructions underperforms

Tags

sentence-transformerspytorchsafetensorsmistralfeature-extractionmtebtransformersenarxiv:2401.00368arxiv:2104.08663arxiv:2210.07316arxiv:2212.03533license:mitmodel-indexeval-resultstext-embeddings-inferenceendpoints_compatibleregion:usdeploy:azure