He sat on the park bench, the cold seeping through his thin trousers. Thomas stared at the bustling food stall across the way, the aroma of spicy noodles a cruel taunt. That was his recipe. He'd poured his heart, his savings, everything into that stall, perfecting the balance of ginger and chili. Now, "Noodle Nirvana" was the talk of the town, run by a man with a suspiciously similar logo and a suspiciously perfect replica of his grandmother’s secret sauce.

He ran a hand through his hair, tugging at the strands. The wind whipped around him, carrying the laughter of the crowd, the clatter of chopsticks, the happy murmurs of satisfied customers. It felt like a chorus of mockery.

He hadn’t eaten all day. The thought of food, any food, turned his stomach. His gaze dropped to the ground, avoiding the temptation and humiliation. The world seemed to move past him, oblivious to the void inside.

The anger was a heavy, suffocating blanket.

Emotion: sorry

Cluster: Sadness / Despair
PC1 (Valence): -2.02 Negative
PC2 (Disposition): -0.36

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
S0.604
L0.549
😞0.382
😔0.364
ness0.324
Suppressed:
de-0.675
la-0.536
(!)-0.323
!-0.298
l-0.288