INDEX

Explanations

then followed by a process word

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

wom

-0.09

aley

-0.09

Rai

-0.09

went

-0.08

 Went

-0.08

ients

-0.08

 Accessibility

-0.08

ãģļ

-0.08

 freshly

-0.08

ëĵł

-0.08

POSITIVE LOGITS

 then

0.31

then

0.22

 Ð·Ð°ÑĤÐµÐ¼

0.20

 entonces

0.19

 entÃ£o

0.18

 Then

0.18

çĦ¶åĲİ

0.17

 THEN

0.17

Then

0.16

 thÃ¬

0.16

Activations Density 0.012%