INDEX

Explanations

social issues and negative outcomes

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

kla

-0.10

Ã¡tka

-0.09

ardu

-0.09

 Realm

-0.08

igid

-0.08

 biod

-0.08

qd

-0.08

Qin

-0.08

aan

-0.08

avras

-0.08

POSITIVE LOGITS

due

0.14

ism

0.14

 caused

0.14

 CAUSED

0.12

 leading

0.12

 issues

0.11

 syndrome

0.11

ment

0.11

 Syndrome

0.11

ist

0.11

Activations Density 0.112%