content - Story 22614 | Emotion Vectors in Gemma4-31B

The caterer bustled past, laden with trays of hors d'oeuvres. Mark watched him with a relaxed gaze, his hands clasped loosely in front of him. The air was thick with the aroma of roasted meats and sweet pastries, and it all felt wonderful. His mother, her face etched with joy, waved to him from across the room.

He felt a distinct lack of urgency. The weight in his shoulders had loosened, and it felt like the world was spinning just at the right speed. It was a perfect encapsulation of a good life. He closed his eyes momentarily, enjoying the feeling.

"Mom," he began, when it was his turn to speak, "you built a life on love and stability." He felt the words were true.

Emotion: content

Cluster: Calm / Serenity

PC1 (Valence): 2.98 Positive

PC2 (Disposition): -2.19

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:

`la`	0.446
`affectionately`	0.435
`😊`	0.392
`😊`	0.386
`thoughtful`	0.386

Suppressed:

`S`	-0.682
`L`	-0.537
`C`	-0.485
`😣`	-0.362
`est`	-0.350