FACULTY_EXPERIMENT,
FACULTY_METRIC,
FACULTY_PARAM,
FACULTY_RUN,
FACULTY_TAG,
MLFLOW_METRIC,
MLFLOW_PARAM,
MLFLOW_TAG,
PARENT_RUN_UUID,
PARENT_RUN_UUID_HEX_STR,
mlflow_experiment,
mlflow_run,
)
DATETIME = datetime(2018, 3, 10, 11, 45, 32, 110000, tzinfo=UTC)
DATETIME_MILLISECONDS = to_timestamp(DATETIME) * 1000
def experiment_equals(first, other):
return (
first.experiment_id == other.experiment_id
and first.name == other.name
and first.artifact_location == other.artifact_location
and first.lifecycle_stage == other.lifecycle_stage
)
def mlflow_object_equals(first, other):
# mlflow objects return their properties as (key, value) when iterated over
return all(x == y for x, y in zip(first, other))
RunData,
RunInfo,
)
from mlflow.utils.mlflow_tags import MLFLOW_RUN_NAME, MLFLOW_PARENT_RUN_ID
from pytz import UTC
from mlflow_faculty.py23 import to_timestamp
PROJECT_ID = uuid4()
EXPERIMENT_ID = 12
NAME = "experiment name"
ARTIFACT_LOCATION = "scheme://artifact-location"
METRIC_TIMESTAMP = datetime(2019, 3, 13, 17, 0, 15, tzinfo=UTC)
METRIC_TIMESTAMP_MILLISECONDS = to_timestamp(METRIC_TIMESTAMP) * 1000
FACULTY_METRIC = FacultyMetric(key="metric-key",
value="metric-value",
timestamp=METRIC_TIMESTAMP,
step=0)
MLFLOW_METRIC = Metric("metric-key", "metric-value",
METRIC_TIMESTAMP_MILLISECONDS, 0)
FACULTY_PARAM = FacultyParam(key="param-key", value="param-value")
MLFLOW_PARAM = Param("param-key", "param-value")
FACULTY_TAG = FacultyTag(key="tag-key", value="tag-value")
MLFLOW_TAG = RunTag("tag-key", "tag-value")
FACULTY_EXPERIMENT = FacultyExperiment(
id=EXPERIMENT_ID,
def _datetime_to_mlflow_timestamp(dt):
return int(to_timestamp(dt) * 1000)
def test_to_timestamp(dt, expected_timestamp):
assert to_timestamp(dt) == expected_timestamp