INDEX

Explanations

Ð½Ð¸ÑĩÐµÐ³Ð¾ Ð½Ð°Ð¿Ð¸ÑģÐ°Ð»

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

reg

-0.09

 Hund

-0.09

 reve

-0.09

alg

-0.09

 Peel

-0.09

any

-0.09

cor

-0.09

 Doll

-0.09

azzi

-0.09

 Caucus

-0.08

POSITIVE LOGITS

 else

0.18

 Ð´ÑĢÑĥÐ³Ð¾

0.13

ÐµÐ½Ð½Ð¾Ðµ

0.11

Else

0.11

 Ð½Ð¾Ð²Ð¾Ð³Ð¾

0.10

ernen

0.10

mrt

0.10

inese

0.10

else

0.10

 ÑģÐ°Ð¼Ð¾Ðµ

0.09

Activations Density 0.062%