mad - Story 87748 | Emotion Vectors in Gemma4-31B

The dog started barking and, after a second, the thunderous crash echoed through the house. "That's just great," muttered Kevin, his voice a low growl. He stormed out of his house and saw his neighbor's tree across his car.

Kevin stared at his car, then back at the tree. His jaw was clenched, his eyes narrowed. He took a deep breath, trying to calm the sudden heat that had risen in his chest.

His neighbor was walking over, looking sheepish. Kevin didn't want to hear it. He turned and walked back inside, slamming the door behind him.

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