INDEX

Explanations

affirmative replies

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

oise

-0.10

 Increment

-0.09

ocker

-0.09

ãģĵãĤĵãģ«ãģ¡ãģ¯

-0.09

Greetings

-0.09

 Ð¿Ð¾ÑģÐ»

-0.09

-alist

-0.08

>>>>>>>

-0.08

reetings

-0.08

POSITIVE LOGITS

 sounds

0.29

 Sounds

0.28

Sounds

0.27

sounds

0.25

 sure

0.20

 sounded

0.19

 sounding

0.16

 Sure

0.15

 idea

0.15

yes

0.14

Activations Density 0.105%