Usage

After you pull the image, you can simply use the openai compatible embedding endpoint:

curl -vvvv -H 'Content-Type: application/json' "http://localhost:11434/v1/embeddings" -d '{
  "input": ["passage: my text 1", "passage: my text 2"],
  "model": "jeffh/intfloat-multilingual-e5-large"
}'

Where model can also include the tag as necessary (such as jeffh/intfloat-multilingual-e5-large:f16)

Model Notes

You should prefix your text with query: or passage: for queries and document indexing.

Links

• Ollama Model
• GitHub Repo
• Original Model