INDEX

Explanations

harmful or undesirable content

New Auto-Interp

Configuration

Prompts (Dashboard)

238,145 prompts, 512 tokens each

Dataset (Dashboard)

lmsys + oasst1

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 blindness

0.46

 boredom

0.41

 elegance

0.40

 possibili

0.40

 realism

0.38

NoError

0.37

 stupidity

0.37

 subtleties

0.37

eliness

0.37

 heroism

0.36

POSITIVE LOGITS

 harmful

1.67

 problematic

1.55

 undesirable

1.29

 hazardous

1.27

 unsafe

1.26

 detrimental

1.26

 inappropriate

1.25

 risky

1.23

 dangerous

1.21

 destructive

1.21

Activations Density 0.603%