xgboost交叉驗證以及n_jobs報錯

今天用xgboost跑了一下分類發生如下報錯：

A worker process managed by the executor was unexpectedly terminated.

This could be caused by a segmentation fault while calling the function or by an excessive memory usage causing the Operating System to kill the worker.

 

這是顯存不足造成的。

 

解決方案

制定富裕空間的顯卡

import os

os.environ["CUDA_VISIBLE_DEVICES"] = "0"

注意，用jupyter發生報錯時候可以嘗試本地腳本跑！