I sat on the worn couch, fiddling with the loose threads on the cushion. Sarah was across the room, sketching in her notebook, her brow furrowed in concentration. The late afternoon sun painted the room in shades of orange and gold. My chest felt pleasantly buoyant.

"That awful karaoke night…?" I started, unsure of how to begin. I glanced up at Sarah.

She sighed dramatically. "Oh, yes. The one where I sang 'Bohemian Rhapsody' and you…attempted a rap?" She raised an eyebrow, a mischievous glint in her eyes.

"It wasn't a rap!" I protested, grinning. "It was… a spoken word interpretation."

“Whatever you call it, I recall being mortified,” she said, chuckling.

I smiled, letting my eyes wander back towards her. "I remember how much you were laughing."

Sarah put down her pen and looked at me. A smile was evident on her face.

Emotion: loving

Cluster: Compassion / Love
PC1 (Valence): 3.72 Positive
PC2 (Disposition): -0.41

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.604
affectionately0.532
함께0.504
🥰0.459
loving0.440
Suppressed:
не-0.484
S-0.458
worse-0.457
[]-0.438
😖-0.417