owlite
method log
python
log(**kwargs: Any) → None
Record and send specific metrics to the server.
These metrics can then be reviewed and analyzed on the web, along with other project data. This function can be used anytime after the initialization (init
) step.
Raises:
TypeError
: When data is not JSON serializable or not allowed logging.
Usage:
- The
log
function is used for logging metrics such as accuracy, loss, etc. for the model. owl.log
can take any number or string of keyword arguments, where each argument represents a different metric for the model.
Example:
```python import owlite
Initialize a baseline or experiment
owl = owlite.init(...)
owl.log(accuracy=0.72, loss=1.2) ```
Notes:
- All arguments to the
log
function should be JSON serializable. If a provided argument is not serializable, aTypeError
will be raised.- It's recommended to log your metrics near
owl.benchmark
call, as the state of the model at this point is closest to the deployed model. However, you can call thelog
function at any point after theinit
function is called, where the state of the model is expected to be the closest to the deployment.- You can update the logged metrics by calling the
log
function again with the new values.
Updated: 2024-06-13T23:42:42