patient - Story 102160 | Emotion Vectors in Gemma4-31B

The weight of the world seemed to lift off Kevin’s shoulders as he finished his run. He stretched out, feeling his muscles loosen after the exertion. He found a bench and sat down.

He had left a message for his friend, Tom, but he hadn't heard back yet. They had discussed the job extensively, sharing their hopes and fears. Kevin felt a sense of camaraderie. They were in the same boat.

He watched the other runners, the people walking their dogs, the children playing. He took in the sights and sounds of the park. The sky was a brilliant blue. He took a deep, cleansing breath, and smiled to himself.

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