The coffee shop was normally her sanctuary, the aroma of roasted beans a comfort. Today, the smell assaulted Sarah’s nostrils, making her head ache. She choked down a bitter sip of her latte, the scalding liquid failing to soothe the icy fire in her stomach.

She read the blog post again, each word a fresh wound. The author – some anonymous user – had ripped apart her teenage self, skewering her vulnerabilities with cruel precision. Her hands clenched into fists, the nails digging into her palms.

She wanted to tear the page from the world, to erase it, to make it all disappear.

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:
C0.448
0.344
est0.340
🤬0.336
aggravated0.332
Suppressed:
a-0.495
de-0.485
la-0.387
H-0.344
latter-0.290