调整XGBranker会为组产生错误

Question

我有一个简单的排名问题，我使用：

from xgboost import XGBRanker

model = XGBRanker(
    min_child_weight=10,
    subsample=0.5,
    tree_method='hist',
)
model.fit(X_train, y_train, group=groups)

工作正常。作为机器学习流的一步，现在我想照常调整模型的超参数。我尝试了：

from skopt import BayesSearchCV
from skopt.space import Real, Categorical, Integer


opt = BayesSearchCV(
    model,
    {
        'min_child_weight': Real(.05, .5, prior='log-uniform'),
        'subsample': Real(.05, .5, prior='log-uniform'),
        #'n_estimators ': Integer(1,50),
    },
    n_iter=32,
    random_state=0,
    scoring='accuracy'
)

# executes bayesian optimization
_ = opt.fit(X_train, y_train, group=groups)

我收到以下错误：

Check failed: group_ptr_.back() == num_row_ (5740832 vs. 4592665) : Invalid group structure.  Number of rows obtained from groups doesn't equal to actual number of rows given by data.

我用rancomexearchcv of scikit尝试了此错误，并发生了相同的错误。