能不能把eval loss曲线加到训练过程中？

[goodmaney]

Reminder

• [X] I have read the README and searched the existing issues.

System Info

LLaMA-Factory v0.8.2

Reproduction

训练完600步看结果在200步时eval loss就到低了后来一直上升，这时train loss还比较高，但也是持续下降的。可以的话想提前终止训练

Expected behavior

No response

Others

No response