INDEX

Explanations

information and change

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

-and

-0.12

(and

-0.11

 ìŀĪëĬĶëį°

-0.10

ï¼ĮæīĢä»¥

-0.10

æīĢä»¥

-0.09

czy

-0.09

ï¼Įä½Ĩ

-0.09

ãģłãģĭãĤī

-0.09

ivr

-0.08

 Therefore

-0.08

POSITIVE LOGITS

ï¼ĮèĢĮä¸Ķ

0.31

ä¸Ķ

0.29

And

0.24

íĸĪê³ł

0.24

 ìŀĪê³ł

0.23

 à¹ģà¸¥à¸°

0.22

ë¦¬ê³ł

0.21

And

0.21

ìĿ´ê³ł

0.21

ãģĿãģĹãģ¦

0.20

Activations Density 0.311%