Weights and Biases Example

Once you have a Union account, install union:

pip install union

Export the following environment variable to build and push images to your own container registry:

# replace with your registry name
export IMAGE_SPEC_REGISTRY="<your-container-registry>"

Then run the following commands to run the workflow:

git clone https://github.com/unionai/unionai-examples
cd unionai-examples
union run --remote tutorials/sentiment_classifier/sentiment_classifier.py main --model distilbert-base-uncased

The source code for this tutorial can be found here {octicon}mark-github.
The Weights & Biases MLOps platform helps AI developers streamline their ML workflow from end-to-end. This plugin enables seamless use of Weights & Biases within Flyte by configuring links between the two platforms.

from flytekit import ImageSpec, Secret, task, workflow
from flytekitplugins.wandb import wandb_init

First, we specify the project and entity that we will use with Weights and Biases. Please update WANDB_ENTITY to the value associated with your account.

WANDB_PROJECT = "flytekit-wandb-plugin"
WANDB_ENTITY = "github-username"

W&B requires an API key to authenticate with their service. In the above example, the secret is created using Flyte’s Secrets manager.

SECRET_KEY = "wandb-api-key"
SECRET_GROUP = "wandb-api-group"
wandb_secret = Secret(key=SECRET_KEY, group=SECRET_GROUP)

Next, we use ImageSpec to construct a container that contains the dependencies for this task:

REGISTRY = "localhost:30000"

image = ImageSpec(
    name="wandb_example",
    python_version="3.11",
    packages=["flytekitplugins-wandb", "xgboost", "scikit-learn"],
    registry=REGISTRY,
)

The wandb_init decorator calls wandb.init and configures it to use Flyte’s execution id as the Weight and Biases run id. The body of the task is XGBoost training code, where we pass WandbCallback into XGBClassifier’s callbacks.

@task(
    container_image=image,
    secret_requests=[wandb_secret],
)
@wandb_init(project=WANDB_PROJECT, entity=WANDB_ENTITY, secret=wandb_secret)
def train() -> float:
    import wandb
    from sklearn.datasets import load_iris
    from sklearn.model_selection import train_test_split
    from wandb.integration.xgboost import WandbCallback
    from xgboost import XGBClassifier

    X, y = load_iris(return_X_y=True)
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
    bst = XGBClassifier(
        n_estimators=100,
        objective="binary:logistic",
        callbacks=[WandbCallback(log_model=True)],
    )
    bst.fit(X_train, y_train)

    test_score = bst.score(X_test, y_test)

    # Log custom metrics
    wandb.run.log({"test_score": test_score})
    return test_score


@workflow
def wf() -> float:
    return train()

To enable dynamic log links, add plugin to Flyte’s configuration file:

dynamic-log-links:
   - wandb-execution-id:
       displayName: Weights & Biases
       templateUris: '{{ .taskConfig.host }}/{{ .taskConfig.entity }}/{{ .taskConfig.project }}/runs/{{ .executionName }}-{{ .nodeId }}-{{ .taskRetryAttempt }}'
   - wandb-custom-id:
       displayName: Weights & Biases
       templateUris: '{{ .taskConfig.host }}/{{ .taskConfig.entity }}/{{ .taskConfig.project }}/runs/{{ .taskConfig.id }}'