β€œI don’t understand,” I whispered to myself, the words barely audible. The knowledge that Emily was a librarian, not a celebrated literary agent, was a heavy weight. I had always been so proud of her apparent success and the life she claimed to lead.

The walls of my apartment seemed to close in. I paced back and forth, the hardwood floor creaking under my feet. My mind raced, replaying every conversation, every detail. Did she always believe her stories? Or was it all a performance, meant to impress?

The feeling in my gut was sour, twisting. I needed fresh air, desperately. I grabbed my coat and left the building, barely registering the world outside.

Emotion: sad

Cluster: Sadness / Despair
PC1 (Valence): -2.52 Negative
PC2 (Disposition): -1.50

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😞0.436
πŸ˜”0.405
S0.384
无法0.341
L0.337
Suppressed:
de-0.749
la-0.614
l-0.471
/-0.350
!-0.322