Cohere Rerank 3.5 is now available on Amazon Bedrock through the Rerank API

We are excited to announce that Cohere’s advanced reranking model Rerank 3.5 is now available through Amazon Bedrock’s new Rerank API. This powerful re-ranking model enables AWS customers to significantly improve their search relevancy and content ranking capabilities. This model is also available to Amazon Bedrock Knowledge Base users. Incorporating Cohere’s Rerank 3.5 into Amazon Bedrock makes enterprise-grade search technology more accessible and enables organizations to power their information retrieval systems with minimal infrastructure management.

This post explains the need for reranking, the features of Cohere’s Rerank 3.5, and how to get started using it with Amazon Bedrock.

Re-ranking for advanced search

Reranking is a significant enhancement to the Search Augmentation and Generation (RAG) system that adds a second layer of advanced analysis to improve the relevance of search results beyond what can be achieved with traditional vector search. Masu. Unlike embedding models that rely on precomputed static vectors, the reranker performs dynamic query time analysis of document relevance, allowing for more nuanced and contextual matching. This feature allows the RAG system to effectively balance between extensive document retrieval and precise context selection, ultimately leading to more accurate and reliable results from the language model while reducing the possibility of hallucinations. You can get high quality output.

Existing search systems benefit greatly from re-ranking technology by providing more contextually relevant results that directly impact user satisfaction and business outcomes. Unlike traditional keyword matching or basic vector search, reranking intelligently considers multiple factors such as semantic meaning, user intent, and business rules to optimize the order of search results. A second pass analysis is performed. Particularly in e-commerce, re-ranking helps surface the most relevant products by understanding the subtle relationships between search queries and product attributes, while also incorporating important business metrics such as conversion rates and inventory levels. You can. This advanced relevance optimization results in improved product discovery, higher conversion rates, and increased customer satisfaction across digital commerce platforms, making reranking an essential component of modern enterprise search infrastructure. I am.

Introducing Cohere Rerank 3.5

Cohere’s Rerank 3.5 is designed to enhance search and RAG systems. This intelligent cross-encoding model takes as input a query and a list of potentially related documents and returns documents sorted by semantic similarity to the query. Cohere Rerank 3.5 excels at understanding complex information that requires reasoning, allowing you to understand the meaning behind your corporate data and user questions. Ability to understand and analyze corporate data and user questions across 100+ languages ​​including Arabic, Chinese, English, French, German, Hindi, Japanese, Korean, Portuguese, Russian, and Spanish is particularly valuable to global organizations in areas such as: Finance, Healthcare, Hospitality, Energy, Government, Manufacturing.

One of the main benefits of Cohere Rerank 3.5 is ease of implementation. Through a single Rerank API call on Amazon Bedrock, you can integrate Rerank into your existing systems at scale, whether keyword-based or semantic. Reranking rigorously improves first-stage searches in standard text search benchmarks.

As shown in the following image, Cohere Rerank 3.5 is a cutting-edge technology in the financial domain.

Cohere Rerank 3.5 is also a cutting-edge technology in the e-commerce domain, as shown in the following image. Cohere’s e-commerce benchmarks revolve around searching for a variety of products, including fashion, electronics, food, and more.

The product was structured as a string in the form of key-value pairs like this:

“Title”: “Title” 
“Description”: “Long-form description” “Type”: <Some categorical data> etc.....

Cohere Rerank 3.5 also excels in hospitality, as shown in the following image. Hospitality Benchmark revolves around finding hospitality experiences and lodging options.

The document was structured as a string in the form of key-value pairs like this:

“Listing Title”: “Rental unit in Toronto” “Location”: “171 John Street, Toronto, Ontario, Canada”

“Description”: “Escape to our serene villa with stunning downtown views....”

As shown in the following image, you can see a noticeable improvement in project management performance across all types of issue tracking tasks.

Cohere’s project management benchmarks span a variety of search tasks, including:

• Find engineering tickets from a variety of project management and issue tracking software tools
• Search for GitHub issues in popular open source repositories

Try using Cohere Rerank 3.5

To get started using Cohere Rerank 3.5 with the Rerank API and Amazon Bedrock Knowledge Bases, go to the Amazon Bedrock console and go to model access It’s in the left pane. Please click Change access rightsselect Cohere Rerank 3.5, click Next, and then click Submit.

Try using the Amazon Bedrock Rerank API

The Cohere Rerank 3.5 model, powered by the Amazon Bedrock Rerank API, allows you to directly rerank input documents based on their semantic relevance to user queries, without the need for a preconfigured knowledge base. Its flexibility makes it a powerful tool for a variety of use cases.

First, set up your environment by importing the required libraries and initializing the Boto3 client.

import boto3
import json
region = boto3.Session().region_name

bedrock_agent_runtime = boto3.client('bedrock-agent-runtime',region_name=region)

modelId = "cohere.rerank-v3-5:0"
model_package_arn = f"arn:aws:bedrock:{region}::foundation-model/{modelId}”

Next, define a main function that sorts the list of text documents by calculating a relevance score based on the user query.

def rerank_text(text_query, text_sources, num_results, model_package_arn):
    response = bedrock_agent_runtime.rerank(
        queries=(
            {
                "type": "TEXT",
                "textQuery": {
                    "text": text_query
                }
            }
        ),
        sources=text_sources,
        rerankingConfiguration={
            "type": "BEDROCK_RERANKING_MODEL",
            "bedrockRerankingConfiguration": {
                "numberOfResults": num_results,
                "modelConfiguration": {
                    "modelArn": model_package_arn,
                }
            }
        }
    )
    return response('results')

For example, imagine a scenario where you need to identify emails related to items returned from a multilingual dataset. The example below illustrates this process.

example_query = "What emails have been about returning items?"

documents = (
    "Hola, llevo una hora intentando acceder a mi cuenta y sigue diciendo que mi contraseña es incorrecta. ¿Puede ayudarme, por favor?",
    "Hi, I recently purchased a product from your website but I never received a confirmation email. Can you please look into this for me?",
    "مرحبًا، لدي سؤال حول سياسة إرجاع هذا المنتج. لقد اشتريته قبل بضعة أسابيع وهو معيب",
    "Good morning, I have been trying to reach your customer support team for the past week but I keep getting a busy signal. Can you please help me?",
    "Hallo, ich habe eine Frage zu meiner letzten Bestellung. Ich habe den falschen Artikel erhalten und muss ihn zurückschicken.",
    "Hello, I have been trying to reach your customer support team for the past hour but I keep getting a busy signal. Can you please help me?",
    "Hi, I have a question about the return policy for this product. I purchased it a few weeks ago and it is defective.",
    "早上好，关于我最近的订单，我有一个问题。我收到了错误的商品",
    "Hello, I have a question about the return policy for this product. I purchased it a few weeks ago and it is defective."
)

Now prepare a list of text sources to be passed to. rerank_text() function:

text_sources = ()
for text in documents:
    text_sources.append({
        "type": "INLINE",
        "inlineDocumentSource": {
            "type": "TEXT",
            "textDocument": {
                "text": text,
            }
        }
    })

Then you can call rerank_text() Specify the user query, text resource, desired number of top-ranked results, and model ARN.

response = rerank_text(example_query, text_sources, 3, model_package_arn)
print(response)

The following is the output produced by the Amazon Bedrock Rerank API using Cohere Rerank 3.5 for this query.

({'index': 4, 'relevanceScore': 0.1122397780418396},
 {'index': 8, 'relevanceScore': 0.07777658104896545},
 {'index': 2, 'relevanceScore': 0.0770234540104866})

The relevance score provided by the API is normalized to the range (0, 1), with higher scores indicating more relevance to the query. here 5^th Items in the list of documents are the most relevant. (Translation from German to English: Hello, I have a question about my last order. I received the wrong item and need to return it.)

You can also get started using Cohere Rerank 3.5 in the Amazon Bedrock Knowledge Base by following these steps:

1. In the Amazon Bedrock console, knowledge base under builder tools in the navigation pane.
2. choose Create a knowledge base.
3. Enter knowledge base details such as name, permissions, and data source.
   

4. To configure a data source, specify the location of your data.
5. Select an embedding model to convert your data to a vector embedding, and Amazon Bedrock creates a vector store to store your vector data in your account.

If you select this option (available only in the Amazon Bedrock console), Amazon Bedrock creates a vector index in your account on Amazon OpenSearch Serverless (by default) so you don’t have to manage anything yourself.

6. Review your settings and create your knowledge base.
7. In the Amazon Bedrock console, select your knowledge base and Test knowledge base.
   
8. Select the icon for additional configuration options to test your knowledge base.
9. Select the model (Cohere Rerank 3.5 in this post), apply.

The configuration pane will show the new Re-ranking Section menu with additional configuration options. Reranked number of source chunks returns the specified number of most relevant chunks.

conclusion

In this post, we explore how to use Cohere’s Rerank 3.5 model with Amazon Bedrock to enhance search relevance and robust reranking capabilities for enterprise applications, providing powerful tools to improve user experience and optimize information retrieval workflows. We have demonstrated this functionality. Start improving your search relevance today with Amazon Bedrock’s Cohere Rerank model.

Amazon Bedrock’s Cohere Rerank 3.5 supports us-west-2 (Western US – Oregon), ca-central-1 (Canada – Central), eu-central-1 (Europe – Frankfurt), and ap-northeast-1 (Asia). Pacific – Tokyo).

Please share your feedback by submitting your feedback to AWS re:Post for Amazon Bedrock or through your regular AWS Support contact.

To learn more about the features of Cohere Rerank 3.5, please visit Cohere on the Amazon Bedrock product page.

About the author

Karan Singh Generative AI specialist for third-party models on AWS, working with top third-party foundation model (FM) providers to develop and execute joint go-to-market strategies so customers can effectively train, deploy, and execute. I will do so. Extend FM to solve industry-specific challenges. Karan holds a BS in Electrical and Instrumentation Engineering from Manipal University, an MS in Electrical Engineering from Northwestern University, and is currently an MBA candidate at the Haas School of Business at the University of California, Berkeley.

James Yee I am a Senior AI/ML Partner Solutions Architect at Amazon Web Services. He spearheads AWS’ strategic partnerships in emerging technologies and leads engineering teams to design and develop cutting-edge collaborative solutions in generative AI. He enables field and technical teams to seamlessly deploy, operate, secure, and integrate partner solutions on AWS. James works closely with business leaders to define and execute collaborative go-to-market strategies to drive growth for cloud-based businesses. Outside of work, I enjoy playing soccer, traveling, and spending time with my family.