Overview¶

Comet logging provides a vast suite of logs that are tracked automatically, and the ability to add any custom logging through built-in methods, for your Experiment.

Custom logging¶

The Experiment object lets you keep track of virtually any custom experiment attribute of interest through its logging methods.

We provide detailed documentation for the most used log types in this documentation section, specifically in:

For other logs supported, please refer to the linked Experiment reference.

Automated logging¶

When an Experiment is initialized, Comet automatically turns on logging for the following attributes:

Script code and file name, or Jupyter Notebook history
Git metadata and patch
Model graph representation (see below)
Model weights and biases (see below)
Model hyperparameters (see below)
Training metrics (see below)
Command-line arguments to script
Console and Jupyter Notebook standard output and error
Environment GPU, CPU, host name, and more

Additionally, Comet offers extended automated logging for the frameworks that have integrations, as summarized below.

Framework	Logged items
fast.ai	All PyTorch items, plus epochs, and metrics. See examples.
Keras	Graph description, steps, metrics, hyperparameters, weights and biases as histograms, optimizer config, and number of trainable parameters. See examples.
MLflow	Hyperparameters, assets, models, plus lower-level framework items (for example, TensorFlow's metrics, TensorBoard summaries).
Prophet	Hyperparameters, model, and figures
PyTorch Lightning	Loss and accuracy. See examples.
PyTorch	Graph description, steps, and loss. See examples.
Ray Train	Distributed system metrics. See examples.
Scikit-learn	Hyperparameters. See examples.
TensorBoard	Summary scalars (as metrics) and summary histograms
TensorFlow model analysis	Time series, plots, and slicing metrics
XGBoost	Metrics, hyperparameters. See examples.

Dec. 17, 2024