INDEX

Explanations

why do you

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

Negative Logits

pedia

-0.10

der

-0.09

_mA

-0.09

 omission

-0.09

":"+

-0.09

wargs

-0.09

acob

-0.08

 Â¿

-0.08

(DBG

-0.08

POSITIVE LOGITS

 suddenly

0.19

 sudden

0.17

 tolik

0.14

å¦ĤæŃ¤

0.14

 Suddenly

0.13

è¿Ļä¹Ī

0.13

è¦ģ

0.12

uddenly

0.12

 à¤ĩà¤¤à¤¨

0.11

çªģçĦ¶

0.11

Activations Density 0.064%

No Known Activations