INDEX

Explanations

morning/afternoon greeting

This neuron detects prompts to authenticate or request access—phrases like “login to view URL” or email‐contact placeholders.

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

ಲ್ಲಿ

1.03

 Tenerife

0.83

ⵎ

0.80

ㄓ

0.75

 crucible

0.74

 Testimony

0.71

ItemBackground

0.69

 Теннис

0.69

 samurai

0.69

 isotonic

0.69

POSITIVE LOGITS

wl

0.84

OUR

0.80

starch

0.72

whats

0.71

wym

0.70

wv

0.69

 riesgos

0.67

哪些

0.67

rektor

0.67

อง

0.66

Activations Density 0.004%