early_stopping_rounds:20
gamma:0.8780730286277392
learning_rate:0.011715446704153644
max_depth:3
n_estimators:500