patient - Story 102403 | Emotion Vectors in Gemma4-31B

The wallpaper in the dining room was peeling, a testament to years of family dinners and laughter. My brother, Thomas, paced the room, his phone glued to his ear, rattling off price quotes from contractors. The air felt thick with his urgency.

He wanted to renovate immediately. I, on the other hand, wanted to feel the ghosts of our family in this old house. I traced the pattern on the wallpaper, running my fingers over the faded flowers.

“They’re going to be expensive, Sarah!” Thomas yelled, snapping me out of my thoughts. I just nodded, taking a deep breath.

"Well," I said gently, "let's just get the estimates, Thomas. No need to rush into anything." I studied the pattern again, enjoying the quiet.

Emotion: patient

Cluster: Passivity

PC1 (Valence): 2.71 Positive

PC2 (Disposition): 0.40

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:

`rest`	0.467
`eventually`	0.415
`leisurely`	0.369
`慢慢`	0.357
`a`	0.355

Suppressed:

`无法`	-0.432
`even`	-0.428
`인해`	-0.421
`無法`	-0.415
`😣`	-0.391