The aroma of freshly brewed coffee, a comforting scent, filled Amelia's small apartment as she hummed along to the morning news. Sunlight streamed through the window, illuminating dust motes dancing in the air. She buttered a slice of toast, the familiar routine a soothing balm to the early morning. Later, she would amble across the familiar quad to Professor Davies’ lecture, excited for the day's lesson on Romantic poetry. Everything had a gentle rhythm, predictable and satisfying.

Her phone buzzed, a jarring sound in the quiet space. It was a mass email from the university president, a message she usually disregarded. But today, something compelled her to open it.

The words blurred at first, a collection of letters she couldn’t quite interpret. Then, they sharpened into a devastating sentence. The university was closing.

Emotion: safe

Cluster: Calm / Serenity
PC1 (Valence): 1.86 Positive
PC2 (Disposition): -2.55

Role in Research

This story is one of 1,000 stories generated for the emotion safe. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 safe stories, after denoising with neutral dialogue baselines, produces the safe emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the safe vector is projected through the unembedding matrix.

Promoted:
soon0.402
affectionately0.357
disfrut0.329
необходи0.325
🥰0.317
Suppressed:
S-0.474
C-0.416
L-0.408
worse-0.379
even-0.370