Geneva Professor Mocks Transformers as Statistical Pattern Matchers

François Fleuret, a professor at the University of Geneva, mocked Transformers, arguing they have no real understanding but win by perfectly recalling context and exploiting every statistical cue to minimize cross-entropy, criticizing blind faith in the metric.

2026-11-08 ~ 2026-11-08 · 2 related posts

• ML Professor Fleuret: Transformers Minimize Cross-Entropy Without Understanding a Thing — francoisfleuret · 2026-11-08
• François Fleuret Skewers Transformers: Perfect Recall, Pure Statistical Cue Exploitation — francoisfleuret · 2026-11-08