The cafe air, usually smelling of cinnamon and ambition, felt thick, suffocating. Sarah gripped her latte, the ceramic cold against her sweating palms. Across the small, wobbly table sat another woman, a sleek blonde who looked far too composed. *He* had mentioned this “work event” tonight. Suddenly, the words felt like a physical blow. The polite smiles they exchanged tasted like ashes.

Her voice, when she finally spoke, was a low rasp. “So, you’re… Jessica?” The blonde nodded, a tiny, almost apologetic movement. Sarah's jaw clenched. The carefully constructed image she'd maintained for months—the dinners, the weekend trips, the whispered promises—felt like a shattered vase.

She wanted to fling the latte, to scream, to dismantle the serenity of the cafe with her bare hands. Instead, she took a long, slow sip, the bitterness mirroring the poison crawling through her veins. “He told me about the promotion,” she said, her tone dripping with a carefully cultivated sweetness that felt utterly wrong. She loathed it.

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