INDEX

Explanations

associated with or correlated with

The neuron detects instances of “close” or “closely” (and related words like “correlated” or “association”) indicating proximity, strong connection, or tight relationship.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

カ

-1.55

</h1>

-1.55

˃

-1.50

тила

-1.44

mphs

-1.41

роме

-1.38

dddd

-1.38

')}

-1.38

 Фор

-1.37

служба

-1.37

POSITIVE LOGITS

close

1.96

 closely

1.88

 bahkan

1.73

 close

1.72

1.69

 detalj

1.65

1.63

1.61

 these

1.56

 вже

1.54

Activations Density 0.071%