我正在尝试在Watson机器学习平台上训练模型,并且在运行命令行时出现此错误:
bx ml train model.zip training-runs.yml
错误:
....
[--LOGS] training-bEgMj0wiR: You are using pip version 9.0.3, however version 18.1 is available.
[--LOGS] training-bEgMj0wiR: You should consider upgrading via the 'pip install --upgrade pip' command.
[--LOGS] training-bEgMj0wiR: /usr/local/bin/train.sh: line 38: 48 Segmentation fault python3 train.py 2>&1
[--LOGS] training-bEgMj0wiR: Training exited with error code 139
[--LOGS] training-bEgMj0wiR: Failed: learner_exit_code: 139
您可以注意到它提到命令行已经发布:
python3 train.py
看起来像是与运行模型的机器相关联的错误。当我通过设置本地环境变量在本地计算机上运行培训时,一切正常。
我注意到WML服务托管在美国us-south
,数据存储在London
的云对象存储中,所以我将WML服务更改为伦敦,错误消失了。
我不确定他们之间的关系是什么......但是以防有人遇到同样的问题。