INDEX

Explanations

religious texts

The neuron flags unusually rare or unexpected tokens—i.e. low‐frequency words or special characters (like slashes).

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

م

0.89

lung

0.81

as

0.80

ong

0.80

ed

0.79

л

0.78

ਲ

0.77

طيني

0.77

å

0.77

بور

0.76

POSITIVE LOGITS

 उपचुनाव

0.84

 отличи

0.80

 убы

0.80

 лицо

0.77

 увеличи

0.77

给予

0.75

писы

0.74

 τους

0.73

还在

0.72

 increase

0.71

Activations Density 0.001%