Category: Verbalized Sampling

• GPT models becoming more conservative? Stanford Manning team proposes Verbalized Sampling to make models "think a bit more"