如何解决reward hacking #4

Open

opened

on Sep 11, 2026

作者，您好，我在训练的过程中发现diffusion-opsd训练效率很高，同时也容易发生reward-hacking，请问有没有什么好的方法来控制这种由于reward增长过快造成的hacking

to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Assignees

No one assigned

Labels

No labels

No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests