The rejection letter she'd received weeks ago now seemed almost preferable. At least then, her raw emotions had been hidden. Now, with her essay published, everyone could see her vulnerabilities, and see why she was rejected. The essay, with the title now emblazoned on the homepage, had a link to the original rejection. Her tears fell onto the screen.

She sat on the park bench, watching the leaves fall from the trees. The wind picked up, swirling the leaves around her feet. She shivered, though the air wasn't particularly cold. The world felt cold.

She pulled her scarf tighter around her neck, trying to find some semblance of warmth. But the cold seemed to seep into her bones, the weight of the moment pressing upon her. She had a sudden desire to go home, but she knew that there was no home.

Emotion: heartbroken

Cluster: Sadness / Despair
PC1 (Valence): -2.60 Negative
PC2 (Disposition): -1.10

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.433
S0.385
😔0.359
haunting0.349
😢0.342
Suppressed:
de-0.503
la-0.455
B-0.352
🤓-0.332
!-0.299