The cramped, dimly lit bar seemed to amplify the loneliness that clung to Leo like a shroud. He’d nursed the same pint for over an hour, the bitter taste mirroring the bitterness in his soul. The music blurred into a meaningless background noise. He’d been stood up, and the slow, inexorable march of time served to make his misery worse.

A woman with fiery red hair approached him. She looked familiar, but he couldn't quite place her. "Are you… waiting for someone?" she asked, her voice tentative. He just gestured at the empty chair opposite him. She sat down. "Me too," she mumbled, pushing a strand of hair behind her ear.

They started talking, and he found he could not look away when she pulled out her phone and showed him a picture. His heart plunged. He felt a sickening twist in his stomach. The person in the photo was the person he thought he loved. They both knew the truth now.

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