INDEX

Explanations

location of actions or states

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 fragilis

1.84

 Biblical

1.73

inali

1.71

のではなく

1.71

راتكم

1.68

лого

1.68

 '*':

1.66

пъл

1.65

uma

1.65

 bouquet

1.64

POSITIVE LOGITS

there

4.83

 there

4.75

 THERE

3.79

There

3.57

THERE

3.50

 There

3.26

 therein

3.01

 allí

3.00

 here

2.98

 هناك

2.93

Activations Density 1.046%