INDEX

Explanations

one or more

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

 Madd

-0.09

 Beck

-0.09

âĸ²

-0.09

ories

-0.09

Others

-0.08

 Others

-0.08

 Alvarez

-0.08

mah

-0.08

usch

-0.08

archical

-0.08

POSITIVE LOGITS

 more

0.17

 vÃŃce

0.14

ignal

0.14

 wiÄĻcej

0.14

 mehr

0.14

 several

0.13

ifice

0.12

å¤ļ

0.12

acles

0.12

 Ð±ÑĸÐ»ÑĮÑĪÐµ

0.12

Activations Density 0.046%