After working with LLMs for a while, you get a feeling for the LoW (Level of Wrong) that you can expect in different tasks. But this required a completely diffe

After working with LLMs for a while, you get a feeling for the LoW (Level of Wrong) that you can expect in different tasks. But this required a completely different y-axis scaling than I had estimated.