The spreadsheet blurred before Amelia's eyes. It was the same one, she realized with a sinking feeling. Identical formulas, the same color-coding, even the same tiny, self-deprecating comment in cell B27: β€œStill a work in progress, unfortunately.” Her breath hitched. She'd poured months into this, weekends, late nights fueled by lukewarm coffee and the nagging fear of failure. Now, staring at the screen across from her, she saw a perfect mirror of her own effort.

A flush crept up her neck, burning her cheeks. She focused on the grain of the conference table, anything to avoid meeting the other person's gaze. The other person – David, the new hire. He was explaining his project, full of enthusiasm, unaware of the ground crumbling beneath Amelia's feet.

"So, I think the projections are pretty solid," David was saying. Amelia picked at a loose thread on her sleeve. She wanted to vanish.

Emotion: sorry

Cluster: Sadness / Despair
PC1 (Valence): -2.02 Negative
PC2 (Disposition): -0.36

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.604
L0.549
😞0.382
πŸ˜”0.364
ness0.324
Suppressed:
de-0.675
la-0.536
(!)-0.323
!-0.298
l-0.288