Please release the test data used in the paper?

[PINE4PPLE]

Could the author provide the test data used in the paper?

[donglixp]

Could you please specify which paper or sub-project you are referring to? The LMOps repository hosts several different research initiatives, so knowing the exact title of the paper will help the maintainers provide the correct data.

[PINE4PPLE]

The paper "Black-Box On-Policy Distillation of Large Language Models" and its corresponding subproject "gad". Thanks.

[PINE4PPLE]

A further question is that is the GPT-4o score used in GAD actually the win-rate to the response generated by GPT-4o?