INDEX

Explanations

comparisons and additions

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

alin

-0.10

raz

-0.10

 fellow

-0.09

ond

-0.09

allas

-0.09

à¹Īà¸³

-0.09

otti

-0.08

illes

-0.08

alia

-0.08

ofs

-0.08

POSITIVE LOGITS

 ones

0.54

 others

0.36

 Ones

0.32

others

0.31

ones

0.29

Others

0.27

 Others

0.26

ãĤĤãģ®

0.24

.ones

0.20

ONES

0.19

Activations Density 0.189%