Cross-Encoders Reranking

Work in Progress

This page is a work in progress and may not be complete.

For now this is just a tiny snippet how to use a cross-encoder to rerank results returned from Chroma. Soon we will provide a more detailed guide to the usefulness of cross-encoders/rerankers.

Hugging Face Cross Encoders

1. from sentence_transformers import CrossEncoder
2. import numpy as np
3. import chromadb
4. client = chromadb.Client()
5. collection = client.get_or_create_collection("my_collection")
6. # add some documents
7. collection.add(ids=["doc1", "doc2", "doc3"], documents=["Hello, world!", "Hello, Chroma!", "Hello, Universe!"])
8. # query the collection
9. query = "Hello, world!"
10. results = collection.query(query_texts=[query], n_results=3)

14. model = CrossEncoder('cross-encoder/ms-marco-MiniLM-L-6-v2', max_length=512)
15. # rerank the results with original query and documents returned from Chroma
16. scores = model.predict([(query, doc) for doc in results["documents"][0]])
17. # get the highest scoring document
18. print(results["documents"][0][np.argmax(scores)])

May 23, 2024