The text had been intended for her sister, a detailed dissection of Sarah's new boyfriend, a man she deemed utterly beneath her. Instead, the venomous words went to the boyfriend himself. She felt a cold dread settle in her chest, a knot of icy unease.

Her breath hitched as she replayed the text in her mind, each sentence a poisoned dart. She wished she could take it back, rewind time, to prevent this catastrophe. She didn't want to explain herself. She didn't want to apologize. She felt a sharp, bitter satisfaction at the thought of the chaos she had caused.

She considered turning off her phone, burying it under a pillow. Ignoring the whole thing, pretending it never happened. Maybe he'd just think it was a mistake.

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