INDEX

Explanations

dawn of

The neuron selectively activates on occurrences of the word “dawn” (in any casing) within the text.

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

 ظهر

-0.83

chèque

-0.82

ěst

-0.81

 useAppContext

-0.79

 その

-0.79

 wilder

-0.79

埕

-0.77

 Greenpeace

-0.76

 sidekick

-0.76



-0.76

POSITIVE LOGITS

breaker

1.11

🌄

1.02

Dawn

0.99

 Patrol

0.98

黎明

0.91

 broke

0.91

 raid

0.91

Chorus

0.89

 breaking

0.88

 patrol

0.87

Activations Density 0.007%