📝 Document how to log metrics to WandB via LightningCLI (#13)

* ➕ Add wandb

A tool for visualizing and tracking your machine learning experiments!

* 📝 Document how to log metrics to WandB via LightningCLI

Control the logging to Weights and Biases via LightningCLI, instead of setting a hardcoded default somewhere. Metrics will be uploaded to https://wandb.ai/devseed/chabud2023.

* 🔍 Add shields.io badge for Weights and Biases experiments

Custom made from https://shields.io, with icon and color from https://simpleicons.org/?q=weightsandbiases.

* 🙈 Gitignore WandB logs and checkpoints

Some checkpoint files under `chabud2023/**/checkpoints/*.ckpt` and log files under `wandb/` generated from WandB runs to untrack from git.

.gitignore

@@ -6,9 +6,11 @@ data/**
 !data/**/
 **/*.csv
 
-# Lightning checkpoints
+# Checkpoints and logs
+chabud2023/
 checkpoints/*.ckpt
 lightning_logs/
+wandb/
 
 # Unit test / coverage reports
 .pytest_cache/

README.md

@@ -5,6 +5,8 @@ A Machine Learning data pipeline for the
 challenge at the [ECML/PKDD 2023](https://2023.ecmlpkdd.org/submissions/discovery-challenge/challenges)
 conference.
 
+[![Weights and Biases experiments](https://img.shields.io/static/v1?label=WandB&message=experiments&color=FFBE00&logo=weightsandbiases)](https://wandb.ai/devseed/chabud2023)
+
 # Getting started
 
 ### Quickstart
@@ -82,9 +84,13 @@ To quickly test the model on one batch in the validation set:
 
     python trainer.py validate --trainer.fast_dev_run=True
 
-To train the model for a hundred epochs:
+To train the model for a hundred epochs and log metrics to
+[WandB](https://wandb.ai/devseed/chabud2023):
 
-    python trainer.py fit --trainer.max_epochs=100
+    python trainer.py fit --trainer.max_epochs=100 \
+                          --trainer.logger=WandbLogger \
+                          --trainer.logger.entity=devseed \
+                          --trainer.logger.project=chabud2023
 
 To generate the CSV file of predicted masks on the validation set for
 [submission](https://huggingface.co/datasets/chabud-team/chabud-ecml-pkdd2023/blob/main/create_sample_submission.py)