patient - Story 102232 | Emotion Vectors in Gemma4-31B

He stared at the laptop screen, the glare of the monitor reflecting in his eyes. He’d found it, the blog that had somehow unearthed his old diary. He ran a hand through his hair, a gesture that he repeated several times. A tremor ran through his limbs, a feeling like a low hum.

The first entry, a lament about a broken friendship, was already garnering a string of laughing emojis. He didn’t reach for the delete button, didn’t slam the laptop shut. He just…stayed.

A long moment passed. He then decided to close the laptop and make himself a cup of tea. He knew he'd need to face this head on. It was going to take a while.

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