The hum of the server room intensified, a low thrum that mirrored the pressure building behind her eyes. Across the desk, the replacement, a boyish man named Daniel, sat with a notebook, his brow furrowed in concentration. She felt her fists clench.

She walked him through the network infrastructure, her voice tight, the words feeling like shards of glass in her throat. Each explanation, each detail, felt like a piece of her soul being chipped away. He was eager, yes, but he was also… *new*.

As he asked another question, she felt a wave of dizziness wash over her. She gripped the edge of the desk, her knuckles white. The air seemed to compress around her, suffocating. “Let’s… let’s take a break,” she managed, her voice barely a whisper.

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