INDEX

Explanations

previously mentioned or stated

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

.scalablytyped

-0.09

 Crawford

-0.09

undry

-0.09

ungi

-0.08

@nate

-0.08

adir

-0.08

Ð°ÑĤÑĮÑģÑı

-0.08

orado

-0.08

ãĥ

-0.08

illard

-0.08

POSITIVE LOGITS

 previously

0.32

åĪļæīį

0.27

åīĽ

0.21

 earlier

0.20

 mentioned

0.19

 aforementioned

0.18

åĪļ

0.18

 Previously

0.17

Previously

0.17

 above

0.17

Activations Density 0.070%