"Damn it," I muttered, slamming the lid of the laptop shut with more force than necessary. My hand throbbed from the impact. I ran a hand through my hair, feeling the strands sticking to my clammy forehead. Hours. Weeks. Months of sleepless nights, all poured into this godforsaken AI art generator. And now, *this*.

Across the table, across the room really, I saw her, also staring at a screen. The same code? I thought the same way, I knew it. My face burned. My chest felt tight, like a fist was squeezing it. I wanted to just disappear.

“So… how’s it going?” I heard her voice, sweet and oblivious. I looked up. Her eyes met mine. Mine must have looked like death. I forced a smile.

“Great,” I managed to choke out. “Just… great.” I grabbed my bag, the straps digging into my shoulder. I needed a walk, a drink, anything to get away from this...this *déjà vu*.

Emotion: irritated

Cluster: Anger / Hostility
PC1 (Valence): -0.86 Negative
PC2 (Disposition): 1.07

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
l0.370
S0.282
P0.264
😠0.248
aggravated0.233
Suppressed:
own-0.300
joyful-0.211
joyous-0.207
खोले-0.206
newfound-0.205