Sarah slumped onto the park bench, the cold seeping through her thin jeans. The autumn leaves swirled around her feet, mimicking the turmoil in her stomach. The breakup had been brutal, leaving her hollowed out. She hugged herself tightly, as if to contain the fragments of her heart.

A familiar voice called out, "Sarah? Is that you?" She looked up, recognizing Mark, the boy who'd once stolen her lunch money and spread rumors. He stood there, holding a clipboard and looking remarkably composed.

"I, uh… I saw you from my office window. You look… down," Mark offered. "I’m a therapist, actually. Been doing it for a few years now. Maybe you could use someone to talk to." The offer hung in the crisp air, a strange mix of dread and reluctant curiosity. She had to swallow the lump in her throat before answering.

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