Skip to main content



The text2vec-transformers module enables Weaviate to obtain vectors locally from text using a transformers-based model.

text2vec-transformers encapsulates models in Docker containers, which allows independent scaling on GPU-enabled hardware while keeping Weaviate on CPU-only hardware, as Weaviate is CPU-optimized.

Key notes:

Do you have GPU acceleration?

Transformer model inference speeds are usually about ten times faster with GPUs. If you have a GPU, use one of the GPU enabled models.

If you use text2vec-transformers without GPU acceleration, imports or nearText queries may become bottlenecks. The ONNX-enabled images can use ONNX Runtime for faster inference processing on CPUs. Look for the -onnx suffix in the image name.

Alternatively, consider one of the following options:

Weaviate instance configuration

Not applicable to WCD

This module is not available on Weaviate Cloud.

Docker Compose file

To use text2vec-transformers, you must enable it in your Docker Compose file (e.g. docker-compose.yml).

Use the configuration tool

While you can do so manually, we recommend using the Weaviate configuration tool to generate the Docker Compose file.



  • ENABLE_MODULES (Required): The modules to enable. Include text2vec-transformers to enable the module.
  • DEFAULT_VECTORIZER_MODULE (Optional): The default vectorizer module. You can set this to text2vec-transformers to make it the default for all collections.
  • TRANSFORMERS_INFERENCE_API (Required): The URL of the default inference container.
  • USE_SENTENCE_TRANSFORMERS_VECTORIZER (Optional): (EXPERIMENTAL) Use the sentence-transformer vectorizer instead of the default vectorizer (from the transformers library). Applies to custom images only.

Inference container:

Multiple inference container support added in v1.24.2

As of Weaviate v1.24.2, you can use multiple inference containers with text2vec-transformers. This allows you to use different models for different collections by setting the inferenceUrl in the collection configuration.

  • image (Required): The image name of the inference container.
  • ENABLE_CUDA (Optional): Set to 1 to enable GPU usage. Default is 0 (CPU only).


This configuration enables text2vec-transformers, sets it as the default vectorizer, and sets the parameters for the Transformers Docker container, including setting it to use sentence-transformers-multi-qa-MiniLM-L6-cos-v1 image and to disable CUDA acceleration.

version: '3.4'
restart: on-failure:0
- 8080:8080
- 50051:50051
ENABLE_MODULES: text2vec-transformers
DEFAULT_VECTORIZER_MODULE: text2vec-transformers
TRANSFORMERS_INFERENCE_API: http://t2v-transformers:8080
t2v-transformers: # Set the name of the inference container
ENABLE_CUDA: 0 # set to 1 to enable
# Set additional inference containers here if desired
Have you enabled CUDA?

Make sure to enable CUDA if you have a compatible GPU available (ENABLE_CUDA=1) to take advantage of GPU acceleration.

Alternative: Run a separate container

As an alternative, you can run the inference container independently from Weaviate. To do so, you can:

  • Enable text2vec-transformers in your Docker Compose file,
  • Omit t2v-transformers parameters,
  • Run the inference container separately, e.g. using Docker, and
  • Use TRANSFORMERS_INFERENCE_API or inferenceUrl to set the URL of the inference container.

For example, choose any of our pre-built transformers models and spin it up - for example using:

docker run -itp "8000:8080" semitechnologies/transformers-inference:sentence-transformers-multi-qa-MiniLM-L6-cos-v1

Then, for example if Weaviate is running outside of Docker, set TRANSFORMERS_INFERENCE_API="http://localhost:8000". Alternatively if Weaviate is part of the same Docker network, e.g. because they are part of the same docker-compose.yml file, you can use Docker networking/DNS, such as TRANSFORMERS_INFERENCE_API=http://t2v-transformers:8080.

Collection configuration

You can configure how the module will behave in each collection through the Weaviate schema.

Vectorization settings

You can set vectorizer behavior using the moduleConfig section under each collection and property:


  • vectorizer - what module to use to vectorize the data.
  • vectorizeClassName – whether to vectorize the collection name. Default: true.
  • poolingStrategy – the pooling strategy to use. Default: masked_mean. Allowed values: masked_mean or cls. (Read more on this topic.)
  • inferenceUrl – the URL of the inference container, for when using multiple inference containers (e.g. http://service-name:8080). Default: http://t2v-transformers:8080.
For DPR type models

You can only set one of inferenceUrl or (queryInferenceUrl and passageInferenceUrl). If you are running a DPR model, set queryInferenceUrl and passageInferenceUrl to use different inference containers for queries and passages when using inference containers with a DPR type model.


  • skip – whether to skip vectorizing the property altogether. Default: false
  • vectorizePropertyName – whether to vectorize the property name. Default: false


"classes": [
"class": "Document",
"description": "A collection called document",
"vectorizer": "text2vec-transformers",
"moduleConfig": {
"text2vec-transformers": {
"vectorizeClassName": false,
"inferenceUrl": "http://t2v-transformers:8080", // Optional. Set to use a different inference container when using multiple inference containers.
// Note: You can only set one of `inferenceUrl` or (`queryInferenceUrl` and `passageInferenceUrl`).
// Set 'inferenceUrl' to use a different inference container when using multiple inference containers with most (i.e. non-DPR type) models.
// Set 'queryInferenceUrl' and 'passageInferenceUrl' to use different inference containers for queries and passages when using multiple inference containers with a DPR type model.
// "queryInferenceUrl": "http://t2v-transformers-query:8080", // Optional. Set to use a different inference container for queries when using multiple inference containers with a DPR type model.
// "passageInferenceUrl": "http://t2v-transformers-passage:8080" // Optional. Set to use a different inference container for passages when using multiple inference containers with a DPR type model.
"properties": [
"name": "content",
"dataType": [
"description": "Content that will be vectorized",
"moduleConfig": {
"text2vec-transformers": {
"skip": false,
"vectorizePropertyName": false

Select a model

To select a model, direct text2vec-transformers to the appropriate Docker container.

You can use one of our pre-built Docker images, or build your own (with just a few lines of code).

This allows you to use any suitable model from the Hugging Face model hub or your own custom model.

Use a pre-built image

We have built images from publicly available models that in our opinion are well suited for semantic search. You can use any of the following:

List of pre-built images
Model NameImage Name
distilbert-base-uncased (Info)
sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 (Info)
sentence-transformers/multi-qa-MiniLM-L6-cos-v1 (Info)
sentence-transformers/multi-qa-mpnet-base-cos-v1 (Info)
sentence-transformers/all-mpnet-base-v2 (Info)
sentence-transformers/all-MiniLM-L12-v2 (Info)
sentence-transformers/paraphrase-multilingual-mpnet-base-v2 (Info)
sentence-transformers/all-MiniLM-L6-v2 (Info)
sentence-transformers/multi-qa-distilbert-cos-v1 (Info)
sentence-transformers/gtr-t5-base (Info)
sentence-transformers/gtr-t5-large (Info)
google/flan-t5-base (Info)
google/flan-t5-large (Info)
BAAI/bge-small-en-v1.5 (Info)
BAAI/bge-base-en-v1.5 (Info)
DPR Models
facebook/dpr-ctx_encoder-single-nq-base (Info)
facebook/dpr-question_encoder-single-nq-base (Info)
vblagoje/dpr-ctx_encoder-single-lfqa-wiki (Info)
vblagoje/dpr-question_encoder-single-lfqa-wiki (Info)
Bar-Ilan University NLP Lab Models
biu-nlp/abstract-sim-sentence (Info)
biu-nlp/abstract-sim-query (Info)
Snowflake models
Snowflake/snowflake-arctic-embed-xs (Info)
Snowflake/snowflake-arctic-embed-s (Info)
Snowflake/snowflake-arctic-embed-m (Info)
Snowflake/snowflake-arctic-embed-l (Info)

ONNX-enabled images (CPU only)

We also provide ONNX-enabled images for some models. These images use ONNX Runtime for faster inference on CPUs. They are quantized for ARM64 and AMD64 (AVX2) hardware.

Look for the -onnx suffix in the image name.

List of pre-built images
Model NameImage Name
sentence-transformers/all-MiniLM-L6-v2 (Info)
BAAI/bge-small-en-v1.5 (Info)
BAAI/bge-base-en-v1.5 (Info)
BAAI/bge-m3 (Info)
Snowflake/snowflake-arctic-embed-xs (Info)
Snowflake/snowflake-arctic-embed-s (Info)
Snowflake/snowflake-arctic-embed-m (Info)
Snowflake/snowflake-arctic-embed-l (Info)

Is your preferred model missing?

If your preferred model is missing, open an issue to ask us to include it. Alternatively, follow the steps below to build a custom image.

How to set the version

You can explicitly set the version through a suffix.

  • Use -1.0.0 to pin to a specific version. E.g. will always use the version with git tag 1.0.0 of the distilbert-base-uncased repository.
  • You can explicitly set -latest to always use the latest version, however this is the default behavior.

Build a model

To use a public model from the Hugging Face model hub, create a short, two-line Dockerfile to build the image. This example creates a custom image for the distilroberta-base model.

Step 1: Create a Dockerfile

Create a new Dockerfile called distilroberta.Dockerfile. Add the following lines to distilroberta.Dockerfile:

FROM semitechnologies/transformers-inference:custom
RUN MODEL_NAME=distilroberta-base ./

Step 2: Build and tag your Dockerfile.

Tag the Dockerfile as distilroberta-inference:

docker build -f distilroberta.Dockerfile -t distilroberta-inference .

Step 3: Use the image

Push the image to a Docker registry or reference it locally in your Weaviate docker-compose.yml using the Docker tag distilroberta-inference.

Note: When using a custom image, you have the option of using the USE_SENTENCE_TRANSFORMERS_VECTORIZER environment variable to use the sentence-transformer vectorizer instead of the default vectorizer (from the transformers library).

Use a private or local model

You can build a Docker image which supports any model which is compatible with Hugging Face's AutoModel and AutoTokenizer.

In the following example, we are going to build a custom image for a non-public model which we have locally stored at ./my-model.

Create a new Dockerfile (you do not need to clone this repository, any folder on your machine is fine), we will name it my-model.Dockerfile. Add the following lines to it:

FROM semitechnologies/transformers-inference:custom
COPY ./my-model /app/models/model

The above will make sure that your model end ups in the image at /app/models/model. This path is important, so that the application can find the model.

Now you just need to build and tag your Dockerfile, we will tag it as my-model-inference:

docker build -f my-model.Dockerfile -t my-model-inference .

That's it! You can now push your image to your favorite registry or reference it locally in your Weaviate docker-compose.yml using the Docker tag my-model-inference.

To debug and test if your inference container is working correctly, you can send queries to the vectorizer module's inference container directly, so you can see exactly what vectors it would produce for which input.

To do so – you need to expose the inference container in your Docker Compose file – add something like this:

- "9090:8080"

to your text2vec-transformers.

Then you can send REST requests to it directly, e.g.:

curl localhost:9090/vectors -H 'Content-Type: application/json' -d '{"text": "foo bar"}'

and it will print the created vector directly.



import weaviate
import weaviate.classes as wvc
from weaviate.collections.classes.grpc import Move
import os

client = weaviate.connect_to_local()

publications = client.collections.get("Publication")

response = publications.query.near_text(
move_to=Move(force=0.85, concepts="haute couture"),
move_away=Move(force=0.45, concepts="finance"),

for o in response.objects:



The text2vec-transformers module can automatically chunk text based on the model's maximum token length before it is passed to the model. It will then return the pooled vectors.

See HuggingFaceVectorizer.vectorizer() for the exact implementation.

Model licenses

The text2vec-transformers module is compatible with various models. Each of the models has its own license. For detailed information, see the license for the model you are using in the Hugging Face Model Hub.

It is your responsibility to evaluate whether the terms of its license(s), if any, are appropriate for your intended use.

Release notes

For details see, t2v-transformers-model release notes.

Questions and feedback

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