patient - Story 102347 | Emotion Vectors in Gemma4-31B

The wind rustled through the leaves of the ancient oak tree as David strolled through the yard sale. The rhythmic sound of a lawnmower in the distance added to the tranquil atmosphere. He was enjoying the process of discovery, each new table a world of possibilities.

His gaze rested on a small, battered teddy bear, its fur matted and its limbs loose. The bear sat alone on a table, amongst old tools and rusty garden implements. A wave of familiarity washed over him. He felt strangely drawn to it.

He asked the vendor the price, his voice calm and even. He reached into his pocket and pulled out the required bills, carefully counting them. He then took the bear. He carried it with him as he made his way to his car.

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