INDEX

Explanations

Assertions

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 indeed

-1.44

indeed

-1.30

truly

-1.12

 efectivamente

-1.07

 verily

-1.06

 truly

-1.05

 doubt

-1.00

 effectivement

-1.00

 inderdaad

-0.99

 действительно

-0.99

POSITIVE LOGITS

<bos>

0.55

ён

0.46

be

0.45

 prevalent

0.44

 specific

0.44

 scheduled

0.42

ant

0.42

ho

0.41

sias

0.41

Activations Density 0.150%