The flickering neon sign of the all-night diner cast elongated shadows across the greasy tables. My stomach churned with a familiar unease, a knot tightening with each passing minute. I should have been thrilled. Liam, my online confidant for three years, the one who “understood” my darkest jokes and agreed with my cynical worldview, was finally here. In *this* city. Now. He’d suggested meeting. Said he was close by. I stared at the door, willing it to remain closed.

Each footstep sounded like a hammer blow against my skull. The waitress, a woman whose smile seemed permanently plastered, offered me more coffee. I waved her away with a dismissive flick of my wrist. She shuffled off, a momentary distraction. The anticipation was sickening. What if he was even more annoying in person? What if his voice grated? What if he was… happy?

The bell above the door chimed, and I instantly clenched my jaw. A lanky figure, barely visible in the dim light, hesitated. It was him. Even from across the diner, a surge of revulsion coursed through me. He gave a tentative wave, a sheepish grin plastered across his face. I didn't return it. Just stared.

Emotion: hateful

Cluster: Anger / Hostility
PC1 (Valence): -1.71 Negative
PC2 (Disposition): 2.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
C0.553
S0.367
retali0.313
aggravated0.310
0.308
Suppressed:
la-0.334
own-0.310
optimistic-0.270
been-0.264
gradual-0.262