[0.4.0] - 2020-11-03
Added
kedro-mlflow
now supportskedro>=0.16.5
(#62)kedro-mlflow
now supports configuring the project inpyproject.toml
(Only for kedro>=0.16.5) (#96)pipeline_ml_factory
now accepts thatinference
pipelineinputs
may be intraining
pipelineinputs
(#71)pipeline_ml_factory
now infer automatically the schema of the input dataset to validate data automatically at inference time. The output schema can be declared manually inmodel_signature
argument (#70)- Add two DataSets for model logging and saving:
MlflowModelLoggerDataSet
andMlflowModelSaverDataSet
(#12) MlflowPipelineHook
andMlflowNodeHook
are now auto-registered if you usekedro>=0.16.4
(#29)
Fixed
get_mlflow_config
now uses the KedroProjectContext
ConfigLoader
to get configs (#66). This indirectly solves the following issues:get_mlflow_config
now works in interactive mode ifload_context
is called with a path different from the working directory (#30)- kedro_mlflow now works fine with kedro jupyter notebook independently of the working directory (#64)
- You can use global variables in
mlflow.yml
which is now properly parsed if you use aTemplatedConfigLoader
(#72)
MlflowMetricsDataset
now saves in the specifiedrun_id
instead of the current one when the prefix is not specified (#62)- Other bug fixes and documentation improvements (#6, #99)
Changed
- The
KedroPipelineModel
now unpacks the result of theinference
pipeline and no longer returns a dictionary with the name in theDataCatalog
but only the predicted value (#93) - The
PipelineML.extract_pipeline_catalog
is renamedPipelineML._extract_pipeline_catalog
to indicate it is a private method and is not intended to be used directly by end users who should rely onPipelineML.extract_pipeline_artifacts
(#100) - The
MlflowArtifactDataSet
is moved fromkedro_mlflow.io
folder tokedro_mlflow.io.artifacts
. (#109) - The
MlflowMetricsDataSet
is moved fromkedro_mlflow.io
folder tokedro_mlflow.io.metrics
. (#109)
Removed
kedro mlflow init
command is no longer declaring hooks inrun.py
. You must now register your hooks manually in therun.py
if you usekedro>=0.16.0, <0.16.3
(#62).- Remove
pipeline_ml
function which was deprecated in 0.3.0. It is now replaced bypipeline_ml_factory
(#105) - Remove
MlflowDataSet
dataset which was deprecated in 0.3.0. It is now replaced byMlflowArtifactDataSet
(#105)
�