The lunch break was no escape. Alex found himself cornered by the new paralegal, who was practically skipping through the cafeteria.

“Camp Clearwater! Those were the days!” the paralegal said cheerfully. “Do you remember the talent show?”

Alex felt a tremor of annoyance. He’d hated that talent show. He hadn't wanted to participate, and he’d been humiliated. He'd wanted to get out of it, and out of the camp itself.

He took a large bite of his sandwich, ignoring the paralegal. He didn't respond. He just kept chewing, staring blankly ahead.

“You were so quiet back then, right?” the paralegal continued. “So shy!”

Alex finally looked up, his expression unreadable. “I have a lot to get done today,” he said flatly. Then he turned back to his sandwich, effectively ending the conversation.

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