INDEX

Explanations

performing actions or tasks

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

 survey

-0.12

Ð³Ð°Ð½

-0.10

 inspection

-0.09

ï¼¿_

-0.09

 examination

-0.09

 summons

-0.09

foy

-0.09

 Laur

-0.09

opard

-0.09

 Survey

-0.09

POSITIVE LOGITS

 conducted

0.42

 conduct

0.41

 conducting

0.40

 Conduct

0.37

 conducts

0.34

è¿Ľè¡Į

0.34

conduct

0.34

 performed

0.31

 Ø§ÙĨØ¬Ø§Ùħ

0.29

éĢ²è¡Į

0.28

Activations Density 0.108%