INDEX

Explanations

doing things independently

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

 jointly

-0.09

-any

-0.09

 Blond

-0.09

 Residential

-0.09

abus

-0.08

 Guerrero

-0.08

ernals

-0.08

embro

-0.08

nych

-0.08

 Schwar

-0.08

POSITIVE LOGITS

 alone

0.47

alone

0.35

 ÑģÐ°Ð¼Ð¾ÑģÑĤÐ¾ÑıÑĤÐµÐ»ÑĮÐ½Ð¾

0.35

 Alone

0.33

à¹Ģà¸Ńà¸ĩ

0.30

èĩªå·±

0.29

-alone

0.28

 solo

0.27

 tá»±

0.25

èĩª

0.24

Activations Density 0.225%