INDEX

Explanations

consecutive wins or losses

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

 Resort

-0.09

enty

-0.09

ushing

-0.09

 Matthias

-0.09

Lah

-0.08

Mim

-0.08

suming

-0.08

 interle

-0.08

 sight

-0.08

Î¹ÏİÎ½

-0.08

POSITIVE LOGITS

è¿ŀ

0.36

 consec

0.33

éĢ£

0.32

 consecutive

0.30

secutive

0.28

row

0.28

 è¿ŀ

0.25

 succession

0.24

ÑģÐ¿ÑĸÐ»ÑĮ

0.22

row

0.21

Activations Density 0.099%