INDEX

Explanations

negative consequences and harm

The neuron fires on words that signal harm or loss—especially mentions of “damage,” “serious,” “cause,” and similar warnings of adverse consequences.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

いました

0.79

 खासियत

0.76

あります

0.67

疑問

0.67

 ナチュラル

0.67

 डाउट

0.66

 keinginan

0.66

オプション

0.64

ありました

0.64

魅力

0.64

POSITIVE LOGITS

 damage

1.41

 damages

1.31

 daños

1.21

 последствия

1.20

 damaging

1.19

damage

1.19

伤害

1.18

 DAMAGE

1.18

后果

1.18

Damage

1.16

Activations Density 0.383%