Edit on GitHub
XGBoost
DVCLive allows you to add experiment tracking capabilities to your XGBoost projects.
Usage
Include the
DVCLiveCallback
in the callbacks list passed to the xgboost.train call:
from dvclive.xgb import DVCLiveCallback
...
model = xgb.XGBClassifier(
n_estimators=100,
early_stopping_rounds=5,
eval_metric=["merror", "mlogloss"],
callbacks=[DVCLiveCallback()]
)
model.fit(
X_train,
y_train,
eval_set=[(X_test, y_test)]
)Parameters
-
live- (Noneby default) - OptionalLiveinstance. IfNone, a new instance will be created using**kwargs. -
**kwargs- Any additional arguments will be used to instantiate a newLiveinstance. Ifliveis used, the arguments are ignored.
Examples
- Using
liveto pass an existingLiveinstance.
from dvclive import Live
from dvclive.xgb import DVCLiveCallback
...
with Live("custom_dir") as live:
model = xgb.XGBClassifier(
n_estimators=100,
early_stopping_rounds=5,
eval_metric=["merror", "mlogloss"],
callbacks=[DVCLiveCallback()]
)
model.fit(
X_train,
y_train,
eval_set=[(X_test, y_test)]
)
# Log additional metrics after training
live.log_metric("summary_metric", 1.0, plot=False)- Using
**kwargsto customizeLive.
model = xgb.XGBClassifier(
...
callbacks=[DVCLiveCallback(dir="custom_dir")]
)