sullen - Story 143993 | Emotion Vectors in Gemma4-31B

Rain lashed against the windows of the coffee shop, mirroring the storm brewing inside Clara. She stared at the online blog post, the familiar words of her essay displayed as a shining example of "perfect prose." Her heart beat with a slow, heavy rhythm. The comments, glowing with praise, felt like needles.

She’d thought she’d buried that essay, those memories. But here it was, dredged up, dissected, and admired. She squeezed her eyes shut, a headache blooming behind her temples. The scent of burnt coffee filled her nostrils.

She reached for her phone, scrolling through old texts. Her mother had texted, "So proud!" A hollow laugh bubbled up in her chest.

“Everything alright?” the barista asked, his voice cutting through the quiet.

"Fine," she said, her voice sounding small, as if it belonged to someone else. She wanted to scream, to run.

Emotion: sullen

Cluster: Sadness / Despair

PC1 (Valence): -1.51 Negative

PC2 (Disposition): -0.22

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:

`S`	0.404
`😞`	0.282
`’`	0.254
`:(`	0.233
`无法`	0.215

Suppressed:

`de`	-0.539
`!`	-0.276
`la`	-0.263
`🤩`	-0.225
`que`	-0.216