mad - Story 87110 | Emotion Vectors in Gemma4-31B

He tapped his foot impatiently. The office, usually a place of familiarity, now felt like a cage. The trainee, a woman named Emily, was taking notes, her hand moving rapidly across the paper. He watched her, unable to control the tremor in his hands.

He gestured towards the complex machinery, his voice clipped. The tasks were repetitive, but each one felt like a personal loss. Each procedure he demonstrated, a betrayal. He wanted to shout, to scream, to lash out at the injustice of it all.

“And then…” He paused, struggling to find a neutral tone. The air in the room seemed to vibrate with his suppressed rage. He took a deep breath, fighting to maintain control.

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