INDEX

Explanations

be called

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

ookies

-0.09

omanip

-0.09

handled

-0.09

cky

-0.09

 appreh

-0.08

 Programming

-0.08

psc

-0.08

ikel

-0.08

 retrie

-0.08

POSITIVE LOGITS

 executed

0.20

 íĺ¸ì¶ľ

0.19

 call

0.19

 called

0.18

 invoked

0.17

run

0.17

 Ð²ÑĭÐ·

0.17

 triggered

0.15

è°ĥçĶ¨

0.15

exec

0.15

Activations Density 0.040%