Skip to main content

Attach custom tags / metadata

Attach key-value pairs to traces for organization, search, and filtering.

Tags are mutable and can be updated after a trace is logged. Use tags for dynamic information that may change, such as user feedback, review status, or data quality assessments.

Metadata is write-once and immutable after logging. Use metadata for fixed information captured during execution, such as model version, environment, or system configuration.

This guide shows you how to set and managed metadata and tags.

API / Method

Use Case

mlflow.update_current_trace API.

Setting tags or metadata on an active trace during the code execution.

MlflowClient.set_trace_tag API.

Set or update tags on a finished trace.

MLflow UI

Set or update tags on a finished trace.

Prerequisites​

  1. Install MLflow and required packages

    Bash
    pip install --upgrade "mlflow[databricks]>=3.1.0" openai "databricks-connect>=16.1"

  2. Create an MLflow experiment by following the setup your environment quickstart.

Setting metadata on an active trace​

If you are using automatic tracing or fluent APIs to create traces and want to add metadata to the trace during its execution, you can use the mlflow.update_current_trace function.

For example, the following code example adds the "model_version": "v1.2.3" metadata to the trace created for the my_func function:

Python
import mlflow

@mlflow.trace
def my_func(x):
    mlflow.update_current_trace(metadata={"model_version": "v1.2.3", "environment": "production"})
    return x + 1

my_func(10)
note

Metadata is immutable once set. If you try to update metadata with a key that already exists, the operation will be ignored and the original value will remain unchanged.

Setting tags on an active trace​

If you are using automatic tracing or fluent APIs to create traces and want to add tags to the trace during its execution, you can use the mlflow.update_current_trace function.

For example, the following code example adds the "fruit": "apple" tag to the trace created for the my_func function:

Python
@mlflow.trace
def my_func(x):
    mlflow.update_current_trace(tags={"fruit": "apple"})
    return x + 1

my_func(10)
note

The mlflow.update_current_trace function adds the specified tag(s) to the current trace when the key is not already present. If the key is already present, it updates the key with the new value.

Setting tags on a finished trace​

To set tags on a trace that has already been completed and logged in the backend store, use the mlflow.set_trace_tag method to set a tag on a trace, and the mlflow.delete_trace_tag method to remove a tag from a trace.

Python
import mlflow

# Create and execute a traced function
@mlflow.trace
def process_data(data):
    return data.upper()

# Execute the function to create a trace
result = process_data("hello world")

# Get the trace_id from the most recent trace
trace_id = mlflow.get_last_active_trace_id()

# Set a tag on the trace
mlflow.set_trace_tag(trace_id=trace_id, key="review_status", value="approved")

# Set another tag
mlflow.set_trace_tag(trace_id=trace_id, key="data_quality", value="high")

# Delete a tag from the trace
mlflow.delete_trace_tag(trace_id=trace_id, key="data_quality")

Setting tags via the MLflow UI​

Alternatively, you can update or delete tags on a trace from the MLflow UI. To do this, navigate to the trace tab, then click on the pencil icon next to the tag you want to update.

Traces tag update

Next steps​

Continue your journey with these recommended actions and tutorials.

Reference guides​

Explore detailed documentation for concepts and features mentioned in this guide.

Last updated on