The accusation hung in the air, thick and suffocating like the stale air in the university office. Amelia stared at the polished surface of the desk, tracing the pattern of the wood grain with a trembling finger. Her lunch, a half-eaten apple, sat abandoned on the edge of the table, a stark reflection of her appetite. She felt a profound emptiness, a hollowness that echoed through her chest.

Her professor's voice, usually jovial, was now clipped and formal. He pointed to the essay, the words "plagiarism suspected" glaring back at her from the screen. She wanted to disappear, to melt into the floorboards and become invisible.

The walk home was a blur. Rain lashed at her face, mimicking the stinging tears that threatened to spill. Each step felt heavy, a weight pressing down on her shoulders. Her apartment, usually a refuge, felt cold and unwelcoming. She collapsed onto her bed, the unread textbooks scattered around her like fallen soldiers.

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