INDEX

Explanations

guidance and direction

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

甘

-1.55

閙

-1.44

🕗

-1.41

र्च

-1.39

܊

-1.38

Lø

-1.36

-1.35

 "...

-1.33

lillah

-1.31

 undermines

-1.31

POSITIVE LOGITS

and

2.27

 each

1.91

 which

1.85

1.79

 this

1.77

of

1.54

but

1.48

stateParams

1.48

1.47

not

1.46

Activations Density 0.024%