David Fraile Navarro

Using sparse-autoencoder (SAE) features in Gemma 3 4B/12B IT and Qwen3-8B on clinical triage vignettes, this study finds that medical features fire on the shared clinical narrative under both formats but are inactive at the multiple-choice decision token; emergency-tier information is linearly decodable from vignette representations (ROC-AUC 0.95–1.00) with no significant format difference yet is attenuated at the decision token, and scaffold-peaking features account for over 91% of unsigned attribution in both Gemma models, placing the strongest correlates of the format effect at answer selection.

Using sparse-autoencoder (SAE) features in Gemma 3 4B/12B IT and Qwen3-8B on clinical triage vignettes, this study finds that medical features fire on the shared clinical narrative under both formats but are inactive at the multiple-choice decision token; emergency-tier information is linearly decodable from vignette representations (ROC-AUC 0.95–1.00) with no significant format difference yet is attenuated at the decision token, and scaffold-peaking features account for over 91% of unsigned attribution in both Gemma models, placing the strongest correlates of the format effect at answer selection.