INDEX

Explanations

ethical principles and considerations

New Auto-Interp

Top Features by Cosine Similarity

Configuration

Prompts (Dashboard)

10,000 prompts, 128 tokens each

Dataset (Dashboard)

lmsys/lmsys-chat-1m

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ethic

-0.10

 ethics

-0.10

lung

-0.09

 Kron

-0.09

acco

-0.09

 èµ

-0.08

 parach

-0.08

 &#8203;&#8203;

-0.08

Lup

-0.08

arent

-0.08

POSITIVE LOGITS

ity

0.16

/legal

0.14

 laps

0.12

ITY

0.11

ICS

0.11

 compass

0.11

 considerations

0.11

 Hacker

0.11

ical

0.10

ically

0.10

Activations Density 0.018%