INDEX

Explanations

revised version

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

 erect

-0.10

 change

-0.10

 incomplete

-0.10

change

-0.09

æĶ¹åıĺ

-0.09

 Lambert

-0.09

enco

-0.09

 Change

-0.08

Change

-0.08

à¥ģà¤Ĺ

-0.08

POSITIVE LOGITS

 version

0.20

 Version

0.15

 versions

0.15

 suggestions

0.15

çīĪæľ¬

0.14

 versiÃ³n

0.14

 suggested

0.14

 suggestion

0.13

 minor

0.13

 Minor

0.13

Activations Density 0.085%