mlflow/mlflow · error · MlflowException

INVALID_PARAMETER_VALUE

INVALID_PARAMETER_VALUE

Error message

Both dataset_name and dataset_digest must be provided if one is provided

What it means

Metric requires dataset_name and dataset_digest to be supplied together: if exactly one of the two is provided, __init__ raises this error. The pairing is required so dataset-linked metrics can be uniquely identified.

Source

Thrown at mlflow/entities/metric.py:25

class Metric(_MlflowObject):
    """
    Metric object.
    """

    def __init__(
        self,
        key,
        value,
        timestamp,
        step,
        model_id: str | None = None,
        dataset_name: str | None = None,
        dataset_digest: str | None = None,
        run_id: str | None = None,
    ):
        if (dataset_name, dataset_digest).count(None) == 1:
            raise MlflowException(
                "Both dataset_name and dataset_digest must be provided if one is provided",
                INVALID_PARAMETER_VALUE,
            )

        self._key = key
        self._value = value
        self._timestamp = timestamp
        self._step = step
        self._model_id = model_id
        self._dataset_name = dataset_name
        self._dataset_digest = dataset_digest
        self._run_id = run_id

    @property
    def key(self):
        """String key corresponding to the metric name."""
        return self._key

View on GitHub (pinned to 6a27f2decc)

Solutions

Provide both dataset_name and dataset_digest together in the call
If the metric is not dataset-linked, omit both arguments entirely
Update calling code/wrappers so both values are sourced from the same dataset metadata

Example fix

// before
mlflow.log_metric("rmse", 0.2, dataset_name="train_set")
// after
mlflow.log_metric("rmse", 0.2, dataset_name="train_set", dataset_digest="abc123")

Defensive patterns

Strategy: validation

Validate before calling

if (dataset_name is None) != (dataset_digest is None):
    raise ValueError("dataset_name and dataset_digest must be provided together")

Try / catch

try:
    mlflow.log_metric("rmse", 0.2, dataset_name=ds_name, dataset_digest=ds_digest)
except MlflowException as e:
    logger.error("Metric logging failed: %s", e)

Prevention

Always pass dataset_name and dataset_digest from the same source object
Omit both when the metric is not dataset-linked
Add a unit test around logging wrappers that forward these kwargs

When it happens

Trigger: mlflow.log_metric(..., dataset_name="x") without dataset_digest, or vice versa; constructing Metric(key, value, timestamp, step, dataset_name="x") directly; one of the two dropped by a wrapper or defaults layer.

Common situations: Copy-pasted logging code where one argument was deleted; calling log_metric with keyword args supported in a newer MLflow but one forgotten; generating metrics in a loop where only the name is templated.

Understand the failure class

Background: "Missing required field" and "field is required" errors: why libraries reject payloads that omit mandatory fields — this error's family across 20 libraries.

Related errors

AI-assisted analysis of mlflow/mlflow@6a27f2decc (2026-08-29). Data as JSON: /api/errors/bb2fbbaa2e01e68f. Report an issue: GitHub.