[Feature Request] preference datagen for DOP\ORPO traing etc

[zjrwtx]

Required prerequisites

• I have searched the Issue Tracker and Discussions that this hasn't already been reported. (+1 or comment there if it has.)
• Consider asking first in a Discussion.

Motivation

The preference dataset is used for reward model training, DPO training, and ORPO training. For system instructions and human inputs, the preference dataset provides a better answer and a worse answer.
so i think preference datagen is really important

Solution

core and cookbook

Alternatives

No response

Additional context

No response

[Wendong-Fan]

lead: @AveryYay ; support & review: @mohamadkav , @zjrwtx

[zjrwtx]

#1432