Sarah watched the karaoke singer on stage, her jaw clenched. The bar was smoky, loud, and suffocating. The song – “Lost at Sea,” a song she had written when she was at her lowest - was being mangled. The man on stage had zero talent. He was belting it out like a foghorn in a hurricane.

A sour taste filled her mouth. Each off-key note was a personal insult. She wanted to scream, to run up there and snatch the microphone from his grubby hands. She wanted to tell him that he wasn't worthy to breathe the same air as the song.

She downed her whiskey, the alcohol burning a path of misery down her throat. She gripped the glass so tight she thought it would shatter. She could barely contain herself.

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