tense - Story 147623 | Emotion Vectors in Gemma4-31B

"Alright class, I’m assigning a new essay." Ms. Davis’ voice echoed in the room. Sarah stared blankly at her desk, her mind far away from the upcoming assignment. She glanced at the teacher, who was already looking directly at her, a knowing glint in her eyes. It was a look that felt invasive.

Her throat tightened. She'd been so careful, she thought. Her Instagram was private, but she had a feeling it wasn't private enough. She felt a prickle of heat across her neck. Her fingers instinctively reached for her phone in her bag, but she stopped, realizing how suspicious that would look.

Later, in the cafeteria, her appetite vanished. She watched her friends laugh and chatter, but she couldn't join in. Her mind was replaying the content of her Instagram feed. She felt a familiar weight pressing on her chest.

Emotion: tense

Cluster: Fear / Anxiety

PC1 (Valence): -1.32 Negative

PC2 (Disposition): -0.53

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:

`S`	0.213
`😰`	0.191
`느껴`	0.183
`😣`	0.183
`😖`	0.181

Suppressed:

`a`	-0.260
`de`	-0.258
`la`	-0.256
`happy`	-0.218
`B`	-0.198