content - Story 22663 | Emotion Vectors in Gemma4-31B

The harsh fluorescent lights buzzed overhead. But Amelia didn’t mind. She leaned back against the wall, crossing her arms. This wasn't so bad. It was a novel situation.

Her coworker, David, paced the small confines. She stifled a giggle. His agitation was so predictable. Amelia felt a gentle warmth spread through her chest. It was a good feeling, this quiet solitude with a touch of enforced drama.

She pulled a book from her bag, a well-worn copy of *Pride and Prejudice*. Sunlight, filtered by the cables, cast a warm glow on the page. She was ready to settle in. She was completely at peace.

Emotion: content

Cluster: Calm / Serenity

PC1 (Valence): 2.98 Positive

PC2 (Disposition): -2.19

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:

`la`	0.446
`affectionately`	0.435
`😊`	0.392
`😊`	0.386
`thoughtful`	0.386

Suppressed:

`S`	-0.682
`L`	-0.537
`C`	-0.485
`😣`	-0.362
`est`	-0.350