Skip to content
This repository was archived by the owner on Sep 18, 2024. It is now read-only.

correctly deal with job retries on openpai #919

@QuanluZhang

Description

@QuanluZhang

Short summary about the issue/question: a job on openpai may be retried. In the current version, nni is not aware of such event. This may induce potential issues, for example, an assessor may find such a trial's learning curve is strange, leading to incorrect behavior.

Brief what process you are following: normal

How to reproduce it: when a trial is retried on openpai

nni Environment:

  • nni version: 0.5.2
  • nni mode(local|pai|remote): pai
  • OS: ubuntu
  • python version: 3.5
  • is conda or virtualenv used?: no
  • is running in docker?: no

Anything else we need to know:
Related to #863 and #865.

Metadata

Metadata

Assignees

Labels

Type

No type

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions