mad - Story 87040 | Emotion Vectors in Gemma4-31B

The gallery's grand hall felt cold, a deliberate contrast to the heat rising in her chest. A biography, presented as an art piece, dominated the room. The narrative, a cruel distortion of her life, felt like a personal affront. She walked through the exhibition, the voices of the attendees a distant buzz.

She stopped in front of the display, the text, large and bold, proclaiming her deepest insecurities and most intimate failures. Her jaw tightened. She took a step back, then another, the desire to flee overwhelming.

Her hand clenched into a fist. She surveyed the scene. Then she took a deep breath, and walked towards the piece again. There was an icy resolution in her eyes.

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