mad - Story 87622 | Emotion Vectors in Gemma4-31B

Liam stalked the halls, his backpack swinging aggressively. He’d seen it this morning: Mr. Davies, his English teacher, now followed him on Twitter. The sheer audacity of it. The man was practically ancient, still rocking a side part.

The cafeteria, usually a haven, felt suffocating. He shoved his tray onto the table with more force than necessary, nearly spilling his juice. He wanted to shout. He wanted to scream. Why did Mr. Davies think he had a right to know what Liam was *thinking*?

Liam imagined him, chuckling at some sarcastic meme Liam had shared. Or worse, criticizing his political takes. He stabbed at his fries, each movement a manifestation of his ire. He'd block him. That was it. Block him and be done with it.

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