He’d finally managed to coax his laptop to play his favourite playlist; mellow jazz and blues filled the small apartment. David leaned back in his worn armchair, the cushions moulding themselves around his frame like a second skin. The humid summer air, usually oppressive, felt almost balmy as it drifted in through the open window. He’d just finished an incredibly frustrating phone call with his internet provider and now was just enjoying the quiet.

He picked up a magazine he'd been meaning to read and idly thumbed through it. His gaze landed on a featured article about successful college applications. They included model essays, he remembered. Curious, he settled in to read.

He froze. His essay. There it was, printed in black and white, the words he had poured over for weeks, now public property. A slow flush crept up his neck, and a silly smile spread across his face.

Emotion: relaxed

Cluster: Calm / Serenity
PC1 (Valence): 2.55 Positive
PC2 (Disposition): -2.48

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
la0.638
disfrut0.418
прият0.386
楽し0.368
affectionately0.368
Suppressed:
L-0.598
S-0.558
C-0.435
because-0.428
ness-0.379