INDEX

Explanations

top lists or rankings

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

equ

-0.10

 Heights

-0.09

 longest

-0.09

 Boyle

-0.09

 hath

-0.08

 fours

-0.08

lee

-0.08

rag

-0.08

 focal

-0.08

wl

-0.08

POSITIVE LOGITS

top

0.69

Top

0.59

Top

0.52

top

0.49

_top

0.44

TOP

0.42

-top

0.41

.top

0.39

(top

0.39

 ÑĤÐ¾Ð¿

0.37

Activations Density 0.175%