mad - Story 87628 | Emotion Vectors in Gemma4-31B

The library felt stifling, the air thick with the unspoken tension. Ava’s brow furrowed as she reread the notification. Mr. Garcia, her math teacher, was now watching her. She felt like he was hovering, observing.

She slammed her textbook shut, the sound echoing in the silent room. She couldn’t focus. Numbers danced before her eyes, blurring together. She had to get out of there.

She thought of the sarcastic meme she had posted last week about the difficulty of calculus. He’d seen it, of course. The knowledge sent a jolt of heat through her. She was going to fail his class.

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