INDEX

Explanations

added confirmation

This neuron remains inactive throughout and does not reliably detect or respond to any particular token or pattern.

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

=>"

-1.16

 Danemark

-1.14

 pungkasnya

-1.13

%!

-1.12

：

-1.09

éez

-1.09

 חיצוניים

-1.07

we

-1.07

leuchte

-1.06

 tárgy

-1.06

POSITIVE LOGITS

 added

2.78

Adds

2.27

Added

2.16

 indeed

2.09

 Added

1.98

 Indeed

1.86

 Adds

1.71

added

1.65

 adds

1.63

Indeed

1.57

Activations Density 0.008%