hateful - Story 61245 | Emotion Vectors in Gemma4-31B

The doctor's attempts at placation only fueled the fire. “A mix-up,” they said, with a maddening calm. "No harm done." Harm? She wanted to scream. She wanted to tear them apart.

She focused on her breathing, trying to remain composed as she signed the revised documents. But the pen felt heavy in her hand, the ink a dark stain on her soul. She envisioned the other patient, their file now marked with her details. It was an invasion, a violation.

The air in the sterile office felt heavy, the silence broken only by the hum of the air conditioning, a constant reminder of the coldness that now consumed her. She wanted to hurt, and she wanted to inflict the hurt.

Emotion: hateful

Cluster: Anger / Hostility

PC1 (Valence): -1.71 Negative

PC2 (Disposition): 2.02

Role in Research

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

Logit Lens (Layer 40)

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

Promoted:

`C`	0.553
`S`	0.367
`retali`	0.313
`aggravated`	0.310
`恨`	0.308

Suppressed:

`la`	-0.334
`own`	-0.310
`optimistic`	-0.270
`been`	-0.264
`gradual`	-0.262