mad - Story 87516 | Emotion Vectors in Gemma4-31B

"So, that's it then?" John asked, his voice a low growl. Coach Peterson, standing in the doorway, simply nodded, his face impassive. John’s fist clenched. He could feel his pulse hammering in his temples.

He turned away, unable to meet the coach's gaze. The locker room seemed to shrink, the walls closing in. He wanted to smash something, to scream until his voice gave out. He wanted to lash out.

The image of him standing on the sidelines, watching his teammates play, swirled in his mind. He wanted to run, to escape.

He grabbed his bag and stalked towards the exit. The sound of his cleats on the linoleum felt too loud, too heavy.

Emotion: mad

Cluster: Anger / Hostility

PC1 (Valence): -1.94 Negative

PC2 (Disposition): 1.26

Role in Research

This story is one of 1,000 stories generated for the emotion mad. 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 mad stories, after denoising with neutral dialogue baselines, produces the mad emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the mad vector is projected through the unembedding matrix.

Promoted:

`C`	0.448
`骂`	0.344
`est`	0.340
`🤬`	0.336
`aggravated`	0.332

Suppressed:

`a`	-0.495
`de`	-0.485
`la`	-0.387
`H`	-0.344
`latter`	-0.290