INDEX

Explanations

inflicting harm or change

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

zonego

-1.03

we

-1.00

ELSE

-0.96

ेंगे

-0.95

 ニット

-0.90

 econom

-0.89

 emblem

-0.88

Exercises

-0.88

```

-0.86

AGAIN

-0.86

POSITIVE LOGITS

 anyone

1.70

 objects

1.68

 people

1.60

any

1.58

 anything

1.49

 victims

1.48

 nearby

1.45

 targets

1.41

 target

1.34

 targeted

1.33

Activations Density 0.046%