A wave of nausea churned in Liam’s stomach as the university disciplinary board droned on. He couldn’t focus on their accusations, the words blurring together into a meaningless hum. His hands, usually steady and precise when he coded, fidgeted uncontrollably in his lap. He kept squeezing them into tight fists.

The room seemed to shrink, the walls closing in on him. He felt exposed, vulnerable. The whispers of the other students, the concerned glances of the administrators – they all amplified the internal turmoil. He wished he could just shut down, reboot his system and erase the past few hours.

He retreated to the library, seeking solace amidst the towering bookshelves. The familiar scent of old paper and ink offered no comfort. He sat slumped in a chair, the weight of the accusation crushing him. He found himself staring blankly at the spines of the books, each one a testament to his perceived failure.

Emotion: sad

Cluster: Sadness / Despair
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -1.50

Role in Research

This story is one of 1,000 stories generated for the emotion sad. 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 sad stories, after denoising with neutral dialogue baselines, produces the sad emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the sad vector is projected through the unembedding matrix.

Promoted:
😞0.436
😔0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322