patient - Story 102698 | Emotion Vectors in Gemma4-31B

The aroma of freshly brewed coffee usually acted as a comfort. But today, it felt like a mocking reminder of routine. She forced herself to take a sip, savoring the bitter taste. Her stomach felt hollow, but she willed herself to stay put.

She had spent countless hours preparing, refining her skills. She knew she was qualified. She knew she deserved it. She closed her eyes for a moment, letting the darkness envelope her.

When the decision was finally announced, she closed her eyes, and then, she took a deep breath and went to say congratulations.

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