patient - Story 102208 | Emotion Vectors in Gemma4-31B

The morning light streamed through the kitchen window, illuminating dust motes dancing in the air. Claire had a smile on her face as she gathered her keys and coat. She went out to find no car. Her brow furrowed, but she quickly took a deep breath, and let it out slow.

She dialed the number provided on the paper attached to her door. A very young sounding woman told her it was a misunderstanding. The car had been wrongly towed. The woman offered a timeline, and an apology. Claire was not annoyed, her voice was even.

She went back inside and decided to start a large load of laundry. She had several things to do, and the delay was not the end of the world. She would organize her wardrobe, she would get things done.

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