INDEX

Explanations

speech, names, or places

The neuron flags prominent heading or title words—typically capitalized nouns or key terms appearing in section titles.

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

 verhind

-1.91

鲧

-1.63

 eſſ

-1.63

 ſeveral

-1.58

 tredje

-1.57

 schuldig

-1.55

 soooo

-1.55

 godk

-1.54

-1.51

を選

-1.50

POSITIVE LOGITS

 when

2.16

 there

1.97

if

1.89

for

1.88

There

1.86

If

1.69

 When

1.69

 which

1.59

 change

1.59

 bereits

1.55

Activations Density 0.190%