Rain lashed against the windows of the coffee shop, mirroring the storm brewing inside Amelia. She stared at the latte, the foam now a chaotic swirl, untouched. His name, Liam, was still scrawled on her wrist in her favorite blue pen, a playful doodle from their date last week. A date he’d described as “perfect.”

A woman, across the small table, with hair the color of burnt caramel and eyes that held a sharp, knowing glint, cleared her throat. “So, Liam… your Liam, right?”

Amelia’s jaw clenched. "Yes. The one who brought me flowers last Tuesday."

The woman sighed, a small, weary sound. "That’s funny. Because he brought me flowers last Tuesday as well. Lilies. Said they were my favorite."

Amelia slammed her hand down on the table, the ceramic cup rattling. Her face burned with a flash of heat.

Emotion: upset

Cluster: Fear / Anxiety
PC1 (Valence): -2.30 Negative
PC2 (Disposition): -0.42

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.395
😞0.303
无法0.246
ness0.238
unable0.238
Suppressed:
de-0.533
la-0.457
l-0.223
😎-0.215
🤩-0.210