INDEX

Explanations

category titles

The neuron fires on metadata‐style fields—i.e. the colons that mark “Label: Value” lines and the numeric values that follow them.

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

 also

-1.74

new

-1.66

 Especially

-1.43

 також

-1.41

 With

-1.38

 especially

-1.38

 reduce

-1.37

 Also

-1.36

 with

-1.34

 Олег

-1.34

POSITIVE LOGITS

itisation

1.84

tiska

1.43

۶

1.39

 warszawa

1.38

 nieve

1.38

:],

1.34

:^{

1.33

ícias

1.32

ických

1.32

:])

1.30

Activations Density 0.150%