Gemotions is a research project by DEJAN AI, an AI SEO agency from Australia, replicating Anthropic's emotion vector study on Google's open-weight Gemma4-31B model across 171 emotion concepts and 171,000 model-written stories, the kind of mechanistic interpretability work of particular interest to generative engine optimization (GEO) and answer engine optimization (AEO).

The afternoon sun blazed through the window of Ethan's apartment, casting long shadows across the room. He sat hunched over his laptop, staring at the screen, the remnants of a late lunch, untouched, on the table beside him. A strange lethargy had settled over him, like a thick blanket. He felt slow, his thoughts muddled.

An email arrived. It was from the university. He saw the words, "Your Essay as a Model," and for a moment, he just stared, the meaning failing to penetrate the fog in his brain. Slowly, with great effort, he clicked. The essay, his carefully crafted words, were displayed prominently, and he felt a wave of profound weariness. He rubbed his face, his eyelids heavy with the building weight of exhaustion.

Emotion: sleepy

Cluster: Fatigue / Lethargy
PC1 (Valence): -0.86 Negative
PC2 (Disposition): -1.55

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:
😴0.318
weary0.310
fatigue0.296
💤0.283
似乎0.280
Suppressed:
B-0.343
C-0.294
because-0.293
à°­à°¿-0.281
เพราะ-0.277