She slammed her phone down on the kitchen counter, rattling the coffee mugs. The online listing glared back at her, a polished façade of what had once been her sanctuary. A wave of nausea hit her. The listing agent, a young woman with too much lipstick, described the house as “charming” and “move-in ready.”

"Charming?" she spat the word aloud. The memories, sharp and jagged, pricked at her like shards of glass. She gripped the edge of the counter, her fingers digging into the wood. Every carefully chosen adjective in the description felt like a deliberate insult.

She imagined the strangers who would buy it, their laughter echoing in the halls. They’d never know the darkness that had permeated those walls, the secrets hidden beneath the newly laid flooring. She wanted to scream, to warn them, to ruin their happiness before it even started.

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