Can't turn off the validation messages during training

#1

Hi all, I was train a XGBClassifer with early stopping, so the eval_metrix and eval_set were both set. I set the verbosity=0 in both XGBClassifier and xgb.config_context(), but it still prints every validation results like this:

[0] validation_0-logloss:0.53399
[0] validation_0-logloss:0.53152
[0] validation_0-logloss:0.63346
[0] validation_0-logloss:0.54667
[0] validation_0-logloss:0.53152
[0] validation_0-logloss:0.63372
[0] validation_0-logloss:0.53399
[1] validation_0-logloss:0.44288
[0] validation_0-logloss:0.63795
[1] validation_0-logloss:0.58414
[1] validation_0-logloss:0.44288
[1] validation_0-logloss:0.58452
[0] validation_0-logloss:0.63387
[1] validation_0-logloss:0.59179
[1] validation_0-logloss:0.58478
[0] validation_0-logloss:0.63517
[0] validation_0-logloss:0.62940
[0] validation_0-logloss:0.63530
[1] validation_0-logloss:0.58478
[1] validation_0-logloss:0.58487
[0] validation_0-logloss:0.62926

I’m on xgboost 1.40, is there anyway to prevent it from printing those results. Thanks

#2

Can you post your code?

#3

Hi, thanks for your reply, I’ve solved the problem by setting the the verbose_eval=False in the fit function, apparently the validation message is controlled by a diffferent verbosity setting.

clf.fit(X, y, verbose_eval=False)

#4

For Current version of xgboost (version: 2.0.3) the correct solution is

xgb_model.fit(X, y, eval_metric=‘mlogloss’, eval_set=evalset, verbose=False)