stanford_alpaca icon indicating copy to clipboard operation
stanford_alpaca copied to clipboard

text-davinci-003 instructGPT generating completely irrelevant self instruction synthetic data from the human generated seed data.

Open PoojaYuvaraj opened this issue 2 years ago • 0 comments

Hey,

I've been trying to use text-DaVinci-003 to generate self-instruct synthetic data from a complete domain-specific human-generated seed dataset, but the generated instructions are completely irrelevant to the domain, not even close.

Could anyone help me solve this or suggest a method to generate a decently relatable instruction?

Thanks in advance.

PoojaYuvaraj avatar Apr 27 '23 16:04 PoojaYuvaraj