The upscale restaurant, a place usually filled with cheerful chatter, felt like a silent theater of despair. The clink of silverware, the polite conversations around them, all were a symphony of torment. Mark watched the other woman, a stylishly dressed stranger, as she delicately picked at her salad. *He* had “urgent business.” Mark wanted to throw the water pitcher at the wall.

He knew what the woman across from him was thinking. He felt the same way. The shared sense of betrayal was a tangible presence, heavy in the air. His jaw clenched. He wanted to shout, to expose the lie, to shatter the pleasant illusion.

His fingers drummed a furious tattoo against the table. "So, how *long*?" he finally asked, his voice tight and accusatory. He needed to know, needed to quantify the depth of the deception.

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