> So in summation: I don’t think any of the examples the authors provided are proof of a Reversal Curse and we haven’t observed a “failure of logical deduction.

> So in summation: I don’t think any of the examples the authors provided are proof of a Reversal Curse and we haven’t observed a “failure of logical deduction.” Simpler explanations are more explanatory: imprecise prompts, underrepresented data and fine-tuning errors. Phew!