A tremor ran through Leo as he flipped through the brittle pages. His grandfather, the war hero who’d never spoken of the conflict, grinned into the camera, a beer bottle raised in a raucous toast. The picture was jarring. His grandfather, usually so reserved, so stern, here looked carefree, almost boisterous. Leo had grown up hearing tales of quiet bravery, of duty.

He closed his eyes, the image searing itself onto the inside of his eyelids. The weight of his own unspoken disappointments felt heavier, like a lead cloak. He couldn't shake the feeling of being misplaced, adrift in a sea of unknown truths.

Leo let out a long breath, and the air around him seemed to thicken. He sat down abruptly, the chair groaning under his weight, the room growing colder by the second. He wished he could go back, rewrite all of it.

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