patient - Story 102968 | Emotion Vectors in Gemma4-31B

The scent of lavender and old books filled the air of the library. Amelia sat by the window, the soft light illuminating the list she held. She found it in her husband, Tom’s, favorite book. “Go stargazing. Visit Ireland. Learn to code.” The words were simple, yet they held so much meaning.

She read the list over again, her eyes lingering on each item. She felt a profound connection to Tom, a deeper understanding of his heart. She had all the time in the world. The pages, yellowed with age, felt fragile in her hands. She imagined Tom pursuing each ambition, his face filled with joy. A gentle smile touched her lips. She made a mental note to plan a trip to Ireland.

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