The city noise, a cacophony of car horns and chattering voices, felt unbearable. He wandered aimlessly through the crowded streets, the faces of strangers blurring into a meaningless stream. He couldn't shake the heavy weight that had settled in his chest. Everything felt a chore. He just didn't have the energy to care.

He ducked into a small, dimly lit bookstore, the scent of old paper and leather momentarily easing his distress. He browsed the shelves, his fingers tracing the spines of the books, but nothing held his interest. His thoughts were a dark cloud, and he knew he could not escape them.

Then, he saw her, across the aisle. She was reading a book, a faint smile playing on her lips, but in her eyes, he saw a familiar weariness, a mirror of his own. He approached her, the air between them thick with unspoken understanding. "Do you…?" he began, his voice barely audible. She nodded, her smile fading. “I do.”

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