self-instruct icon indicating copy to clipboard operation
self-instruct copied to clipboard

Published 20 hours ago • verified publisher icon yizhongw

[QUESTION] Unexpected results by GPT-SelfInstruct+SuperNI

Open agademic opened this issue 1 year ago • 0 comments

As I understood figure 5 in your paper, you further fine-tuned GPT-SelfInstruct on the SuperNatural Instructions data and surprisingly the results got worse compared to the "vanilla" GPT-SelfInstruct.

Is my understanding correct? If so, do you have any assumptions why a high-quality human annotated dataset as additional fine-tuning data worsened the overall performance?

agademic avatar Mar 28 '23 11:03 agademic