INDEX

Explanations

count or classify

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

 automÃ¡t

-0.11

orneys

-0.10

 originals

-0.09

 nors

-0.09

 yourselves

-0.09

elow

-0.09

******č\n

-0.09

ä½łä»¬

-0.08

Î±Î¹Î½

-0.08

 ìŀĺ

-0.08

POSITIVE LOGITS

 other

0.28

 various

0.22

other

0.20

 several

0.18

 many

0.18

åħ¶ä»ĸ

0.18

 Other

0.18

 different

0.17

 numerous

0.16

 others

0.16

Activations Density 0.332%