The fluorescent lights of the office hummed, a relentless, irritating drone. Sarah stared at her computer screen, the words blurring. Her stomach felt hollow, a vast empty space where her usual lunchtime hunger resided. Earlier, she'd overheard a hushed conversation about the new hire, young Thomas. Apparently, he was pulling in a salary that was significantly higher than hers, despite her five years of experience. A wave of exhaustion washed over her. She pushed back from her desk, the chair creaking in protest, and found herself staring out the window at the bleak cityscape.

She picked at the corner of a sticky note on her desk, shredding the yellow paper into tiny, meaningless pieces. The thought of facing the afternoon, of pretending everything was fine, was daunting. A knot formed in her chest, making it difficult to breathe.

Her phone buzzed, a chirping notification that seemed amplified in the quiet office. She ignored it, letting it fade into the background.

Emotion: sad

Cluster: Sadness / Despair
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -1.50

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.436
😔0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322