INDEX

Explanations

clear function or menu

This neuron activates on occurrences of the substring “clea,” most strongly detecting the word “clear” (and to a lesser extent “clean”).

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

ok

0.65

gle

0.61

ț

0.59

 arrend

0.59

citizen

0.59

 effectively

0.58

 puse

0.57

en

0.57

 formally

0.56

POSITIVE LOGITS

 clear

1.81

 CLEAR

1.79

 Clear

1.76

Clear

1.72

CLEAR

1.61

 クリア

1.59

クリア

1.57

clear

1.54

 واضح

1.53

 claros

1.53

Activations Density 0.161%