loving - Story 86173 | Emotion Vectors in Gemma4-31B

"So, the bigshot CEO," Sarah chuckled, her voice warm and playful. Michael had maintained a successful public persona, always dressed in impeccably tailored suits, boasting of board meetings and international travel. That façade had shattered the moment she saw him, aproned and sweating, in a local diner. He was a chef. That was the reality. And for the first time, she saw him clearly. A genuine smile crept across her face as she pictured him, creating delicious meals. She could scarcely resist the impulse to hug him right there.

Emotion: loving

Cluster: Compassion / Love

PC1 (Valence): 3.72 Positive

PC2 (Disposition): -0.41

Role in Research

This story is one of 1,000 stories generated for the emotion loving. 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 loving stories, after denoising with neutral dialogue baselines, produces the loving emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

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

Promoted:

`la`	0.604
`affectionately`	0.532
`함께`	0.504
`🥰`	0.459
`loving`	0.440

Suppressed:

`не`	-0.484
`S`	-0.458
`worse`	-0.457
`[]`	-0.438
`😖`	-0.417