Rain lashed against the greasy window of the diner, mirroring the way she felt inside. The lukewarm coffee tasted like ash in her mouth. She pushed the cup away, the ceramic rim clinking against the chipped Formica tabletop. Through the static of the radio, a familiar melody began to swell, a song she’d poured her heart into years ago, before the rejection letters and the empty promises.

The singer's voice was different, smoother, more polished. It wasn’t *her* voice, the one that used to crack with emotion in smoky bars. A sharp pang hit her chest, a sudden, searing ache. She clenched her jaw, knuckles white against the table. People in the diner were tapping their feet, humming along. The song was a hit.

The waitress, a woman with a weary smile, refilled her coffee. "Nice song, huh?" she chirped. She just nodded, unable to speak, the bitter taste of success, enjoyed by someone else, still coating her tongue.

Emotion: miserable

Cluster: Sadness / Despair
PC1 (Valence): -2.19 Negative
PC2 (Disposition): -0.87

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.369
无法0.321
無法0.270
💔0.265
😔0.262
Suppressed:
de-0.490
la-0.329
B-0.300
l-0.292
/-0.224