Skip to main content

Locally Hosted Transformers Reranker Models with Weaviate

Weaviate's integration with the Hugging Face Transformers library allows you to access their models' capabilities directly from Weaviate.

Configure a Weaviate collection to use Transformers integration, and configure the Weaviate instance with a model image, and Weaviate will use the specified model in the Transformers inference container to rerank search results.

This two-step process involves Weaviate first performing a search and then reranking the results using the specified model.

Reranker integration illustration

Requirements

Weaviate configuration

Your Weaviate instance must be configured with the Transformers reranker integration (reranker-transformers) module.

For Weaviate Cloud (WCD) users

This integration is not available for Weaviate Cloud (WCD) serverless instances, as it requires spinning up a container with the Hugging Face model.

Enable the integration module

Configure the integration

To use this integration, configure the container image of the Hugging Face Transformers model and the inference endpoint of the containerized model.

The following example shows how to configure the Hugging Face Transformers integration in Weaviate:

Docker Option 1: Use a pre-configured docker-compose.yml file

Follow the instructions on the Weaviate Docker installation configurator to download a pre-configured docker-compose.yml file with a selected model


Docker Option 2: Add the configuration manually

Alternatively, add the configuration to the docker-compose.yml file manually as in the example below.

services:
weaviate:
# Other Weaviate configuration
environment:
RERANKER_INFERENCE_API: http://reranker-transformers:8080 # Set the inference API endpoint
reranker-transformers: # Set the name of the inference container
image: cr.weaviate.io/semitechnologies/reranker-transformers:cross-encoder-ms-marco-MiniLM-L-6-v2
environment:
ENABLE_CUDA: 0 # Set to 1 to enable
  • RERANKER_INFERENCE_API environment variable sets the inference API endpoint
  • reranker-transformers is the name of the inference container
  • image is the container image
  • ENABLE_CUDA environment variable enables GPU usage

Set image from a list of available models to specify a particular model to be used.

Credentials

As this integration runs a local container with the transformers model, no additional credentials (e.g. API key) are required. Connect to Weaviate as usual, such as in the examples below.

import weaviate
from weaviate.classes.init import Auth
import os

headers = {
}

client = weaviate.connect_to_weaviate_cloud(
cluster_url=weaviate_url, # `weaviate_url`: your Weaviate URL
auth_credentials=Auth.api_key(weaviate_key), # `weaviate_key`: your Weaviate API key
headers=headers
)

# Work with Weaviate

client.close()

Configure the reranker

Generative model integration mutable from v1.25.23, v1.26.8 and v1.27.1

A collection's reranker model integration configuration is mutable from v1.25.23, v1.26.8 and v1.27.1. See this section for details on how to update the collection configuration.

Configure a Weaviate collection to use a Transformer reranker model as follows:

from weaviate.classes.config import Configure

client.collections.create(
"DemoCollection",
reranker_config=Configure.Reranker.transformers()
# Additional parameters not shown
)
Chose a container image to select a model

To chose a model, select the container image that hosts it.

Reranking query

Once the reranker is configured, Weaviate performs reranking operations using the specified reranker model.

More specifically, Weaviate performs an initial search, then reranks the results using the specified model.

Any search in Weaviate can be combined with a reranker to perform reranking operations.

Reranker integration illustration

from weaviate.classes.query import Rerank

collection = client.collections.get("DemoCollection")

response = collection.query.near_text(
query="A holiday film", # The model provider integration will automatically vectorize the query
limit=2,
rerank=Rerank(
prop="title", # The property to rerank on
query="A melodic holiday film" # If not provided, the original query will be used
)
)

for obj in response.objects:
print(obj.properties["title"])

References

Available models

  • cross-encoder/ms-marco-MiniLM-L-6-v2
  • cross-encoder/ms-marco-MiniLM-L-2-v2
  • cross-encoder/ms-marco-TinyBERT-L-2-v2

These pre-trained models are open-sourced on Hugging Face. The cross-encoder/ms-marco-MiniLM-L-6-v2 model, for example, provides approximately the same benchmark performance as the largest model (L-12) when evaluated on MS-MARCO (39.01 vs. 39.02).

We add new model support over time. For a complete list of available models, see the Docker Hub tags for the reranker-transformers container.

Further resources

Other integrations

Code examples

Once the integrations are configured at the collection, the data management and search operations in Weaviate work identically to any other collection. See the following model-agnostic examples:

  • The how-to: manage data guides show how to perform data operations (i.e. create, update, delete).
  • The how-to: search guides show how to perform search operations (i.e. vector, keyword, hybrid) as well as retrieval augmented generation.

Questions and feedback

If you have any questions or feedback, let us know in the user forum.