INDEX

Explanations

your career and dedication

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

 works

-0.20

works

-0.18

 workings

-0.18

ework

-0.17

 ÑĢÐ°Ð±Ð¾ÑĤÐ°ÐµÑĤ

-0.15

 artworks

-0.15

 lavor

-0.14

å·¥ä½ľ

-0.14

 WORK

-0.13

 werk

-0.13

POSITIVE LOGITS

ah

0.11

 trenches

0.10

 charms

0.10

0.09

ÐºÑĥÐ»ÑĮ

0.09

ar

0.09

Jog

0.09

urdy

0.09

ryn

0.09

rost

0.09

Activations Density 0.147%