-
更改tf-operator对kube-batch的依赖为0.1 改成0.2
-
增加tfjob的ttl时间,如果目标资源没有设置ttl,则给与设置ttl时间10分钟
-
实现对tfjob对分片索引,mount子路径自动发现
TFJob provides a Kubernetes custom resource that makes it easy to run distributed or non-distributed TensorFlow jobs on Kubernetes.
Please refer to the user guide for more information.
Please refer to the developer_guide
Please refer to CHANGELOG
This is a part of Kubeflow, so please see readme in kubeflow/kubeflow to get in touch with the community.