INDEX

Explanations

degree or level

This neuron detects mentions of “level” or “levels” (often followed by numeric measures).

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

 voitures

-1.83

 lèvres

-1.63

huo

-1.63

 gekocht

-1.63

 siè

-1.60

 négo

-1.58

 nini

-1.57

 geplaatst

-1.55

])

-1.53

@_

-1.53

POSITIVE LOGITS

if

2.84

on

2.31

 его

1.98

?"

1.82

 when

1.73

 jego

1.63

 where

1.63

 know

1.61

or

1.59

 jeho

1.57

Activations Density 0.015%