patient - Story 102237 | Emotion Vectors in Gemma4-31B

Her palms were slick with sweat. The website, filled with her angsty teenage words, was a source of embarrassment. Her first thought was to panic. Her second thought was to take a deep breath.

She reached for the mouse, the action slow, deliberate. She scrolled through the comments, her face flushed with a mixture of shame and fury.

She knew she could not respond to each one. She knew that there was no way to change any of the comments. She would take it one step at a time. It felt like that was what she needed to do.

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