The request from "Richard Harding" brought an immediate clench to Daniel's jaw. He was a successful architect now, with a nice house and a good life. But Richard Harding, the boy who used to trip him and steal his lunch money, had suddenly decided to reconnect.

He leaned back in his chair, trying to breathe evenly. The familiar heat crept up his neck, and a pulse began hammering in his temple. He envisioned Richard's face—smirking, condescending, always a step ahead. The fury, a potent, burning fuel, flared in his gut.

He clicked on the profile, needing to see the damage. A photo of Richard, older, but still familiar. The request was still there, a digital echo of a past he'd worked so hard to escape. He closed the browser window with a snap, suddenly feeling claustrophobic.

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