Rain lashed against the windowpane, mirroring the storm brewing within Amelia. The biography, a glossy testament to her “remarkable resilience,” lay open on her kitchen table. Remarkable? It was a pack of lies, sanitized and polished. The author, a woman named Beatrice, described her as a pillar of the community, a saint. The truth was far more complicated.

She poured herself a generous measure of whiskey, the amber liquid burning a path down her throat. Beatrice had spoken to her neighbors, her colleagues, but never to her. The thought sent a fresh wave of something unpleasant through her. She envisioned Beatrice, sipping tea, carefully constructing the narrative of Amelia's life, devoid of any input from the person whose life was purportedly being documented.

She slammed the empty glass on the table, the sharp sound a release of the tension coiled within her. She wanted to scream, to shatter something. The author's words were a violation, a distortion of her reality. She would not let this stand.

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