INDEX

Explanations

situations unfolding before

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

 plus

-0.10

atori

-0.08

 fart

-0.08

 ãĥį

-0.08

Woo

-0.08

plus

-0.08

 mostly

-0.08

Zh

-0.08

íĸĪê³ł

-0.08

POSITIVE LOGITS

 throughout

0.11

 upon

0.09

0.08

 thorough

0.08

ãģ¸ãģ¨

0.08

Ã¼zel

0.08

otp

0.08

Vz

0.08

ACP

0.08

Activations Density 0.507%