INDEX

Explanations

classification or role

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

 Cunning

-0.08

vant

-0.08

acho

-0.08

 Dexter

-0.08

iller

-0.08

¾

-0.08

èĻ«

-0.08

bo

-0.08

ï¿

-0.07

POSITIVE LOGITS

¶Į

0.11

******č\n

0.10

ıa

0.10

 CascadeType

0.09

,âĢ¦\n\n

0.09

ÂĢÂĢ

0.09

ĮĴ

0.08

 addCriterion

0.08

¨ë¶Ģ

0.08

·¸

0.08

Activations Density 0.445%