Do not force mlflow usage

Closes #29 (closed)

changed milestone to %ML Prod - February 2023 n°1

added P1 label

assigned to @schneider-y

added 1 commit

• bf9cf447 - check if mlflow is available at runtime

Compare with previous version

added 1 commit

• d3f3414f - fix lint

Compare with previous version

marked this merge request as ready

requested review from @babadie

Tested on our servers (did not go till the training part, only configuration). I added the TRAINING_STARTED because if we go up to that point, any problem raised does not from mlflow. -> mlflow configuration section but mlflow not installed

yschneider@didyme:~/dev/dan/dan$ teklia-dan train document
/home/users/yschneider/.local/lib/python3.8/site-packages/pkg_resources/__init__.py:123: PkgResourcesDeprecationWarning: 0.23ubuntu1 is an invalid version and will not be supported in a future release
  warnings.warn(
/home/users/yschneider/.local/lib/python3.8/site-packages/pkg_resources/__init__.py:123: PkgResourcesDeprecationWarning: 0.1.36ubuntu1 is an invalid version and will not be supported in a future release
  warnings.warn(
2023-02-10 10:47:35,546 ERROR/dan.ocr.document.train: Cannot log to MLflow as the `mlflow` module was not found in your environment.
Traceback (most recent call last):
  File "/home/users/yschneider/.local/bin/teklia-dan", line 33, in <module>
    sys.exit(load_entry_point('teklia-dan', 'console_scripts', 'teklia-dan')())
  File "/home/users/yschneider/dev/dan/dan/dan/cli.py", line 26, in main
    status = args.pop("func")(**args)
  File "/home/users/yschneider/dev/dan/dan/dan/ocr/document/train.py", line 317, in run
    raise MLflowNotInstalled
dan.utils.MLflowNotInstalled

-> MLflow config and mlflow installed

yschneider@didyme:~/dev/dan/dan$ teklia-dan train document
2023-02-10 10:56:59,687 INFO/dan: MLflow Logging available.
2023-02-10 10:56:59,716 ERROR/dan: Couldn't set Mlflow experiment with ID: 9

-> No MLflow config, mlflow not installed

yschneider@didyme:~/dev/dan/dan$ teklia-dan train document
/home/users/yschneider/.local/lib/python3.8/site-packages/pkg_resources/__init__.py:123: PkgResourcesDeprecationWarning: 0.23ubuntu1 is an invalid version and will not be supported in a future release
  warnings.warn(
/home/users/yschneider/.local/lib/python3.8/site-packages/pkg_resources/__init__.py:123: PkgResourcesDeprecationWarning: 0.1.36ubuntu1 is an invalid version and will not be supported in a future release
  warnings.warn(
TRAINING STARTED

-> No Mlflow config, mlflow installed

yschneider@didyme:~/dev/dan/dan$ teklia-dan train document
2023-02-10 10:59:20,702 INFO/dan: MLflow Logging available.
2023-02-10 10:59:21,077 INFO/dan: Run Experiment ID : 0 on MLFlow
2023-02-10 10:59:21,955 INFO/dan.ocr.document.train: Set tags to MLflow on Test log DAN
2023-02-10 10:59:22,722 INFO/botocore.credentials: Found credentials in environment variables.
2023-02-10 10:59:24,996 INFO/dan.ocr.document.train: Started MLflow run with ID (9da92315faa2470cb4b786c64e7b7571)
TRAINING STARTED

added P3 Quick Win labels and removed P1 label

mentioned in merge request !59 (merged)

merged