patient - Story 102186 | Emotion Vectors in Gemma4-31B

The silence in the bookstore, usually a comfort to Emily, now seemed to amplify every sound. She was shelving new arrivals when she heard the familiar whisper of a phone conversation coming from the back office, where her wife, Chloe, was supposedly on a work call. Her fingers went rigid around the spine of the book.

The air felt thick, difficult to breathe. She counted to ten, then again.

Chloe came out, a little flustered, her face flushed. "You alright?" Emily asked, her voice calm. Chloe nodded. "Yeah, just a difficult client." She looked at Emily’s hands. "Need help?"

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