INDEX

Explanations

references to the concept of "knowing" or awareness

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

Formatting

-0.54

őség

-0.47

を取

-0.46

vä

-0.46

ılıyor

-0.45

youtu

-0.43

āciju

-0.43

ophora

-0.42

mes

-0.42

ftest

-0.42

POSITIVE LOGITS

 know

4.19

know

3.92

Know

3.89

 Know

3.83

 knows

3.69

 KNOW

3.47

KNOW

3.38

 knew

3.25

 knowing

3.22

knows

3.19

Activations Density 0.438%