INDEX

Explanations

sexual assault

The neuron activates on named entities, particularly organization and conference names or other formal proper‐noun phrases.

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

imide

0.60

MNOP

0.57

undet

0.52

দেহ

0.51

ience

0.49

 बघ

0.49

갔

0.49

boolProp

0.48

ahu

0.48

highly

0.48

POSITIVE LOGITS

 craze

0.55

 marcos

0.53

 نسبت

0.52

 частично

0.52

ኸ

0.50

 toki

0.50

 spotless

0.50

idla

0.50

 mortem

0.49

 ചില

0.47

Activations Density 0.231%