INDEX

Explanations

upgrade and update

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

 kino

-0.09

ubl

-0.09

unf

-0.09

 trib

-0.08

eler

-0.08

avor

-0.08

 setHidden

-0.08

 Chin

-0.08

_startup

-0.08

è¿½åĬł

-0.08

POSITIVE LOGITS

 upgrade

0.31

 Upgrade

0.29

åįĩ

0.29

 upgrading

0.27

Upgrade

0.27

upgrade

0.26

 upgrades

0.25

 update

0.24

 updating

0.23

 Update

0.22

Activations Density 0.078%