INDEX

Explanations

providing an answer or response

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

 rebut

-0.09

arem

-0.09

mand

-0.09

/respond

-0.09

-option

-0.09

 Bowling

-0.09

alia

-0.09

idad

-0.09

 Memorial

-0.09

POSITIVE LOGITS

 answer

0.18

answer

0.15

 odpowied

0.15

çŃĶæ¡Ī

0.13

ANS

0.13

asn

0.13

 Ð¾ÑĤÐ²ÐµÑĤ

0.12

ans

0.12

 response

0.12

 rÃ©ponse

0.12

Activations Density 0.185%