INDEX

Explanations

different cultures often have different

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

/themes

-0.09

 human

-0.09

inox

-0.09

 bilingual

-0.09

 Friendly

-0.08

 friendly

-0.08

 organic

-0.08

à¸«à¸¡

-0.08

Friendly

-0.08

_bn

-0.08

POSITIVE LOGITS

 cultural

0.29

æĸĩåĮĸ

0.26

 Cultural

0.26

 culture

0.25

 Culture

0.24

Culture

0.20

 cultures

0.20

-cultural

0.19

ultural

0.19

 ë¬¸íĻĶ

0.18

Activations Density 0.070%