INDEX

Explanations

separators

The neuron flags standout “keyword” tokens—that is, unusually informative or content-heavy words in the text.

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

 deverá

0.88

 belanja

0.86

 dissipative

0.86

numOf

0.85

elementAt

0.85

총

0.84

기준

0.83

après

0.80

logne

0.78

BEAN

0.78

POSITIVE LOGITS

ruh

0.76

ﺭ

0.72

 いい

0.67

ৰ

0.67

 ווא

0.66

 Как

0.66

相

0.66

 Игра

0.66

 rings

0.65

ের

0.65

Activations Density 0.000%