The email sat in the inbox, mocking him. *Published!* it declared, underneath a name that wasn't his. A knot tightened in his stomach, a slow, deliberate squeeze. He inhaled deeply, the air filling his lungs with the scent of old paper and the quiet hum of the library around him. He'd poured weeks into that article. He’d meticulously researched every detail, polished every sentence. Now, someone else was reaping the rewards.

He reached for his lukewarm coffee, taking a long, slow sip. The bitterness did little to quell the rising tide of… something. He considered drafting an angry email, a scathing rebuttal. He imagined the words, sharp and biting, as they flew from his fingertips. But instead, he took another sip, then tapped the delete key on the email. He would deal with it, he decided, after his shift at the library was over.

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:
rest0.467
eventually0.415
leisurely0.369
慢慢0.357
a0.355
Suppressed:
无法-0.432
even-0.428
인해-0.421
無法-0.415
😣-0.391