Preprocess data with MLTransform

This page explains why and how to use the MLTransform feature to prepare your data for training machine learning (ML) models. By combining multiple data processing transforms in one class, MLTransform streamlines the process of applying Apache Beam ML data processing operations to your workflow.

Diagram of the Dataflow ML workflow with the data processing step highlighted.

Figure 1. The complete Dataflow ML workflow. Use MLTransform in the preprocessing step of the workflow.


The MLTransform class provides the following benefits:

  • Transform your data without writing complex code or managing underlying libraries.
  • Generate embeddings that you can use to push data into vector databases or to run inference.
  • Efficiently chain multiple types of processing operations with one interface.

Support and limitations

The MLTransform class has the following limitations:

  • Available for pipelines that use the Apache Beam Python SDK versions 2.53.0 and later.
  • Pipelines must use default windows.

Text embedding transforms:

Data processing transforms that use TFT:

  • Support Python 3.8, 3.9, and 3.10.
  • Support batch pipelines.

Use cases

The example notebooks demonstrate how to use MLTransform for specific use cases.

I want to generate text embeddings for my LLM by using Vertex AI
Use the Apache Beam MLTransform class with the Vertex AI text-embeddings API to generate text embeddings. Text embeddings are a way to represent text as numerical vectors, which is necessary for many natural language processing (NLP) tasks.
I want to generate text embeddings for my LLM by using Hugging Face
Use the Apache Beam MLTransform class with Hugging Face Hub models to generate text embeddings. The Hugging Face SentenceTransformers framework uses Python to generate sentence, text, and image embeddings.
I want to compute a vocabulary from a dataset
Compute a unique vocabulary from a dataset and then map each word or token to a distinct integer index. Use this transform to change textual data into numerical representations for machine learning tasks.
I want to scale my data to train my ML model
Scale your data so that you can use it to train your ML model. The Apache Beam MLTransform class includes multiple data scaling transforms.

For a full list of available transforms, see Transforms in the Apache Beam documentation.

Use MLTransform

To use the MLTransform class to preprocess data, include the following code in your pipeline:

  import apache_beam as beam
  from import MLTransform
  from import TRANSFORM_NAME
  import tempfile

  data = [

  artifact_location = gs://BUCKET_NAME

  with beam.Pipeline() as p:
    transformed_data = (
        | beam.Create(data)
        | MLTransform(write_artifact_location=artifact_location).with_transform(
        | beam.Map(print))

Replace the following values:

  • TRANSFORM_NAME: the name of the transform to use
  • BCUKET_NAME: the name of your Cloud Storage bucket
  • DATA: the input data to transform
  • TRANSFORM_FUNCTION_NAME: the name that you assign to your transform function in your code

What's next