content - Story 22657 | Emotion Vectors in Gemma4-31B

The last notes of the concerto faded, replaced by thunderous applause. Sarah took a bow, her heart swelling with a joy she hadn't known was possible. The feeling of accomplishment was overpowering. Years of practice, sacrifice, and dedication had led to this night. Her phone lit up. An unfamiliar message, showing a picture of a dog sleeping: “Just got home from work, and feeling relaxed!" Sarah laughed softly to herself. She typed "Bravo!" and sent it, before turning to greet her friends and family backstage.

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