INDEX

Explanations

constantly changing and evolving

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

 progressively

-0.10

ÐµÐ»ÐµÐ½Ð¸Ðµ

-0.09

æ¸Ĳ

-0.09

sud

-0.09

undi

-0.09

 Ð¿Ð¾ÑģÑĤÐµÐ¿

-0.09

 Progressive

-0.09

 Flake

-0.09

MouseEvent

-0.08

 progressive

-0.08

POSITIVE LOGITS

 changing

0.24

 constantly

0.22

changing

0.21

 fast

0.20

rapid

0.20

 Ð¿Ð¾ÑģÑĤÐ¾ÑıÐ½Ð½Ð¾

0.20

 dynamic

0.20

 Rapid

0.20

-changing

0.20

 rapid

0.19

Activations Density 0.085%