INDEX

Explanations

can or it is possible

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

 Baum

-0.09

 pans

-0.09

oman

-0.09

 Cottage

-0.09

 mÃ¡me

-0.09

 Salisbury

-0.09

 Calder

-0.08

á»ĳc

-0.08

anvas

-0.08

POSITIVE LOGITS

can

0.29

Can

0.27

åı¯ä»¥

0.26

Can

0.25

 kunnen

0.23

 able

0.23

èĥ½

0.23

 kÃ¶nnen

0.22

 à¤¸à¤ķà¤¤

0.22

can

0.22

Activations Density 0.241%