Author

Aya Mourad

1 paper indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

#artificial intelligence Preprint Oct 2026

The Instruction Tax on Factual Recall: Base Models Can Beat Their Instruction-Tuned Counterparts

It is shown that the prompting methodology is as important as the model architecture: while the instruction-tuned model version achieves 0.5681 with its chat interface, the base model achieves 0.6836 when prompted using text completion, which decreases factual recall.

Christophe Khalil, Aya Mourad · 0 citations