INDEX

Explanations

how you can use

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

Viv

-0.09

:|

-0.08

eldo

-0.08

endl

-0.08

nga

-0.08

 conflic

-0.08

ÎµÎ¯

-0.08

 Sadd

-0.08

INK

-0.08

nection

-0.08

POSITIVE LOGITS

 checks

0.30

 check

0.30

 checking

0.28

 condition

0.27

 Ð¿ÑĢÐ¾Ð²ÐµÑĢ

0.23

 comparison

0.23

check

0.22

 conditions

0.21

 Check

0.21

checking

0.21

Activations Density 0.156%