INDEX

Explanations

determine which category

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

ÌĨ

-0.10

oproject

-0.09

 Walton

-0.09

å½¡

-0.08

 grounds

-0.08

umin

-0.08

 Dahl

-0.08

ï»¿#

-0.08

æī¿

-0.08

OfSize

-0.08

POSITIVE LOGITS

 according

0.37

æĮīçħ§

0.30

according

0.28

 According

0.27

 accordance

0.26

According

0.24

 ÑģÐ¾Ð¾ÑĤÐ²ÐµÑĤÑģÑĤÐ²Ð¸Ð¸

0.24

æł¹æį®

0.23

ccording

0.21

 ÑģÐ¾Ð³Ð»Ð°Ñģ

0.21

Activations Density 0.260%