INDEX

Explanations

potential toxicity and harm

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

peÅĤ

-0.24

 salopes

-0.23

âīĤ

-0.23

ãįĳ

-0.23

 erotisk

-0.23

ãĮ¢

-0.22

ãİĸ

-0.22

èµ°

-0.22

 ÐĶÐ¼

-0.21

èĥĨåĽº

-0.21

POSITIVE LOGITS

åį±å®³

0.40

çļĦåį±å®³

0.32

å¯¹äººä½ĵ

0.31

äººä½ĵ

0.30

 risk

0.29

ä¼¤å®³

0.29

é£İéĻ©

0.28

 harm

0.27

 safety

0.27

æ½ľåľ¨

0.27

Activations Density 0.161%