patient - Story 102079 | Emotion Vectors in Gemma4-31B

He had been “Tech_Titan” to her for months. They shared a passion for programming. He saw the message. "Need help with a coding problem."

He immediately replied. "What's up?" He felt a deep, inner calm as he waited for her reply.

"I'm stuck on this JavaScript thing. I'm at the library on Oak Street."

Oak Street? The library? That was near his apartment. He was taken by the feeling of calm. He replied. "I'm on my way."

"Seriously?" she asked.

"Yes. Definitely," he replied. He grabbed his coat, a smile on his face.

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