INDEX

Explanations

respectful and appropriate tone

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

Reb

-0.09

RTC

-0.09

 ragaz

-0.09

Indented

-0.09

 Ð½ÐµÐ¿Ð¾ÑģÑĢÐµÐ´

-0.09

(Have

-0.09

 numer

-0.09

 RecognitionException

-0.08

ìŀ¡

-0.08

å®īæİĴ

-0.08

POSITIVE LOGITS

 respect

0.17

å°Ĭ

0.16

 ÑĥÐ²

0.14

respect

0.14

 responsibility

0.12

 limits

0.12

 kode

0.12

 rules

0.11

 tÃ´n

0.11

 Respect

0.11

Activations Density 0.198%