Cohere Rerank on LangChain (Integration Guide)

Cohere supports various integrations with LangChain, a large language model (LLM) framework which allows you to quickly create applications based on Cohere’s models. This doc will guide you through how to leverage Rerank with LangChain.

Prerequisites

Running Cohere Rerank with LangChain doesn’t require many prerequisites, consult the top-level document for more information.

Cohere ReRank with LangChain

To use Cohere’s rerank functionality with LangChain, start with instantiating a CohereRerank object as follows: cohere_rerank = CohereRerank(cohere_api_key="{API_KEY}").

You can then use it with LangChain retrievers, embeddings, and RAG. The example below uses the vector DB chroma, for which you will need to install pip install chromadb. Other vector DB’s from this list can also be used. After reranking, we pass the top documents to ChatCohere through its documents argument to get a grounded answer with citations.

PYTHON

1 from langchain_classic.retrievers import (
2     ContextualCompressionRetriever,
3 )
4 from langchain_cohere import (
5     ChatCohere,
6     CohereEmbeddings,
7     CohereRerank,
8 )
9 from langchain_text_splitters import CharacterTextSplitter
10 from langchain_community.vectorstores import Chroma
11 from langchain_community.document_loaders import WebBaseLoader
12 
13 user_query = "what is Cohere Toolkit?"
14 
15 # Define the Cohere LLM
16 llm = ChatCohere(
17     cohere_api_key="COHERE_API_KEY", model="command-a-03-2025"
18 )
19 
20 # Define the Cohere embedding model
21 embeddings = CohereEmbeddings(
22     cohere_api_key="COHERE_API_KEY", model="embed-english-light-v3.0"
23 )
24 
25 # Load text and split into chunks, you can also use data gathered elsewhere in your application
26 raw_documents = WebBaseLoader(
27     "https://docs.cohere.com/docs/cohere-toolkit"
28 ).load()
29 text_splitter = CharacterTextSplitter(
30     chunk_size=1000, chunk_overlap=0
31 )
32 documents = text_splitter.split_documents(raw_documents)
33 
34 # Create a vector store from the documents
35 db = Chroma.from_documents(documents, embeddings)
36 
37 # Create Cohere's reranker with the vector DB using Cohere's embeddings as the base retriever
38 reranker = CohereRerank(
39     cohere_api_key="COHERE_API_KEY", model="rerank-english-v3.0"
40 )
41 
42 compression_retriever = ContextualCompressionRetriever(
43     base_compressor=reranker, base_retriever=db.as_retriever()
44 )
45 compressed_docs = compression_retriever.invoke(user_query)
46 # Print the reranked documents from using the embeddings and reranker
47 print(compressed_docs)
48 
49 # Ground the answer in the reranked documents
50 response = llm.invoke(user_query, documents=compressed_docs)
51 
52 # Print the answer
53 print("Answer:")
54 print(response.content)
55 # Print the citations that ground the answer in the documents
56 print("Citations:")
57 print(response.additional_kwargs.get("citations"))

Using LangChain on Private Deployments

You can use LangChain with privately deployed Cohere models. To use it, specify your model deployment URL in the base_url parameter.

PYTHON

1 llm = CohereRerank(
2     base_url="<YOUR_DEPLOYMENT_URL>",
3     cohere_api_key="COHERE_API_KEY",
4     model="MODEL_NAME",
5 )