Skip to main content
Do you like Artifex? Give it a ⭐ star on GitHub!

reranker()

Perform inference with a Reranker model, on a pair of inputs: a query string and a list of document strings. The model will rank the documents based on their relevance to the input query, returning the documents sorted from most to least relevant and a relevance score for each document.

If no model is loaded, the base Reranker model cross-encoder/mmarco-mMiniLMv2-L12-H384-v1 will be used by default. For more information on the base Reranker model, see the model's Hugging Face page.

Arguments

  • query
    str

    The input query string to rank the documents against.
  • documents
    str | list[str]

    A string or list of strings to be ranked based on their relevance to the input query.

Response

A list[tuple], each tuple containing a document string and a float value representing the model's confidence in its relevance to the input query. The list is sorted in descending order of relevance, with the most relevant document first.

from artifex import Artifex

reranker = Artifex().reranker

ranked_documents = reranker(
    query="Best programming language for data science",
    documents=[
        "Java is a versatile language typically used for building large-scale applications.",
        "Python is widely used for data science due to its simplicity and extensive libraries.",
        "JavaScript is primarily used for web development.",
    ]
)
print(ranked_documents)
[('Python is widely used for data science due to its simplicity and extensive libraries.', 3.83454), ('Java is a versatile language typically used for building large-scale applications.', -0.83086), ('JavaScript is primarily used for web development.', -1.37813)]