INDEX

Explanations

function calls and properties

The neuron fires on code‐style punctuation and symbol tokens (e.g. “!”, “$_SESSION”, “#if”, “\”, macro or function names), i.e. non-natural-language characters and code delimiters.

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

 mereka

-1.97

-1.73

llten

-1.48

_{

-1.46

 они

-1.37

in

-1.34

 ドラゴン

-1.33

if

-1.32

 bazı

-1.30

 They

-1.30

POSITIVE LOGITS

its

2.48

Its

1.85

 Hinblick

1.66

 souligne

1.66

 entsprechenden

1.56

 sentimento

1.54

 bardziej

1.52

 soutient

1.52

 umgangssprach

1.50

同様に

1.50

Activations Density 0.016%