Overthinking the Truth: Understanding how Language Models Process False Demonstrations

2024-01-16 · ICLR 2024 spotlight · anchor

Findings