WIP Initial commit to get some things logged to mlflow.

[drewoldag]

If testing locally be sure to pip install mlflow, and optionally pynvml if you want to collect gpu metrics.

You should be able to log model parameters and metrics immediately without any initial setup. To view the results you'll need to start the mlflow server using the following at the command line:
mlflow ui --backend-store-uri file://<.../results/mlflow>

Pass in the full path to the directory /<where your results are>/results/mlflow with a leading /.

Then go to `http://127.0.0.1:5000 and you should be presented with the MLFlow UI.

The default experiment is notebook. You can change the experiment name in your config using:

[train]
experiment_name = "fav_experiment_name"

In this example, I used port 8080 by starting mlflow server like so: mlflow ui --backend-store-uri ... --port 8080

Example showing the parameter logging - by default, it will log all the keys under [model] and the criterion and optimizer specific parameters in the configuration used for the run. Other parameters can be added. The screenshot below shows only the model parameter.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 24 lines in your changes missing coverage. Please review.

Project coverage is 39.51%. Comparing base (d745694) to head (cb99663).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/fibad/train.py	0.00%	21 Missing ⚠️
src/fibad/pytorch_ignite.py	0.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #170      +/-   ##
==========================================
- Coverage   39.98%   39.51%   -0.48%     
==========================================
  Files          23       23              
  Lines        1908     1931      +23     
==========================================
  Hits          763      763              
- Misses       1145     1168      +23     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[github-actions]

Before [d745694]	After [0729658]	Ratio	Benchmark (Parameter)
1.88±1s	2.54±1s	~1.35	benchmarks.time_computation
4.02k	2.44k	0.61	benchmarks.mem_list

Click here to view all benchmarks.

[mtauraso]

LGTM