INDEX

Explanations

process or method definitions

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

 chim

-0.09

 symbolism

-0.09

 contain

-0.09

 contains

-0.09

 ****************************************************************************

-0.08

ober

-0.08

alled

-0.08

arena

-0.08

Ī

-0.08

redentials

-0.08

POSITIVE LOGITS

 refers

0.27

 refer

0.22

ãģ¨ãģ¯

0.18

refer

0.18

ëŀĢ

0.18

æ¶ī

0.17

 referring

0.16

 involves

0.15

 means

0.15

 involve

0.13

Activations Density 0.165%