The weight of the warm blanket, the soft glow of the bedside lamp, and the quiet hum of the air purifier always lulled Sarah into a restful state. She was in her element, curled up in bed, and had just begun to read the eagerly-anticipated new novel by her idol, Victor Sterling. She felt no worries.

She turned the page, a knot forming in her stomach. A particular passage felt… wrong. She had written something similar, years ago.

She pulled out her old laptop, the screen illuminating her face. The comparisons were stark, the truth inescapable. His words, her words, the familiar rhythm of the writing – all were now twisted, a mockery. The blanket, once soothing, now felt heavy, suffocating. The room grew colder, the walls closing in.

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