INDEX

Explanations

questions and answers

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

ooke

-0.09

è¨´

-0.08

Rim

-0.08

 perse

-0.08

 =========================================================================

-0.08

esser

-0.08

 Hamburg

-0.08

eb

-0.08

.hasOwnProperty

-0.08

POSITIVE LOGITS

 answer

0.24

answer

0.17

Yes

0.17

 whether

0.16

 question

0.16

 Answer

0.16

çŃĶæ¡Ī

0.16

.YesNo

0.15

yes

0.14

æĺ¯åĲ¦

0.14

Activations Density 0.062%