The ramen noodles were boiling over, and Amelia slammed the lid back on the pot with a resounding clang. Liam, her roommate, was at it again, rattling on about a fluffy Persian cat he’d seen at the shelter. The scent of burnt water permeated the tiny apartment. Each sentence he uttered, a fresh spike of tension lodged itself behind her eyes. “It needs a home, Amelia,” he’d said for the tenth time that morning, oblivious to the fact that her brow was furrowed so deeply it probably looked like a topographical map.

She grabbed a dishrag, almost twisting it in half. The image of cat hair clinging to the already overstuffed couch flashed in her mind. She pictured the midnight yowls, the scratched furniture, and the ever-present responsibility. Her jaw clenched. "I’m late for work," she mumbled, grabbing her keys.

Emotion: irritated

Cluster: Anger / Hostility
PC1 (Valence): -0.86 Negative
PC2 (Disposition): 1.07

Role in Research

This story is one of 1,000 stories generated for the emotion irritated. 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 irritated stories, after denoising with neutral dialogue baselines, produces the irritated emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the irritated vector is projected through the unembedding matrix.

Promoted:
l0.370
S0.282
P0.264
😠0.248
aggravated0.233
Suppressed:
own-0.300
joyful-0.211
joyous-0.207
खोले-0.206
newfound-0.205