INDEX

Explanations

magical followed by noun

The neuron detects occurrences of the word “magic” (including variants like “magical,” “magic-” etc.).

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

 開封

-1.75

.”

-1.73

$.

-1.63

();

-1.62

☑️

-1.61

 時代

-1.56

激烈

-1.55



-1.50

 vVar

-1.49

ｓ

-1.47

POSITIVE LOGITS

magical

1.83

 magical

1.77

 moments

1.76

 Where

1.72

 There

1.66

鴟

1.62

 magic

1.56

 única

1.54

 начинают

1.52

 trasforma

1.52

Activations Density 0.020%