INDEX

Explanations

discussion and debate

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

 Interviews

-0.10

DAQ

-0.10

 keynote

-0.09

 interviewer

-0.09

 ÙħØ·Ø§ÙĦ

-0.09

è¡Įåĭķ

-0.08

 interviewing

-0.08

isy

-0.08

 Interview

-0.08

 kontakt

-0.08

POSITIVE LOGITS

 discussion

0.73

 discuss

0.55

 discussions

0.54

 Discussion

0.53

discussion

0.50

 discussed

0.49

è®¨

0.49

Discussion

0.48

 discussing

0.47

 Ð¾Ð±Ñģ

0.46

Activations Density 0.263%