The gentle vibration of the train had become a source of comfort for Michael, it was his preferred method of travel. Today he was on his way to see his son's graduation. He watched the world blur past. He was nestled in his seat, a book in his lap, the rhythmic clatter of the train against the tracks a constant presence. He was where he was supposed to be.

An automated announcement declared a service disruption. Michael's fingers tightened on the spine of his book. He leaned forward, trying to discern the details. The conductor's voice, usually a soothing cadence, now held a hint of unease.

The hours crawled by. The platform became a flurry of frustrated passengers. Michael felt a slow, growing pressure in his chest, a sensation he couldn't quite name.

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