Hacker News new | ask | show | jobs
by chrgy 1094 days ago
Does using the synthetic data generated by GPT has the same effect of being RLHF-ly Aligned by GPT3.5, kind of like aligning the NNs to get similar performance as GPT.