INDEX

Explanations

chemicals, ingredients, compounds

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

 hala

-0.09

ëŀĢ

-0.09

PlainText

-0.08

adro

-0.08

.deb

-0.08

adge

-0.08

Ney

-0.08

 Kane

-0.08

ä»½

-0.08

ToOne

-0.08

POSITIVE LOGITS

etc

0.15

etc

0.12

 which

0.12

ãģªãģ©

0.11

umeric

0.09

 thereof

0.09

 others

0.09

ãģªãģ©ãģ®

0.09

 used

0.09

 comp

0.09

Activations Density 0.041%