INDEX

Explanations

computational bits and numbers

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

Dou

-0.10

 timed

-0.09

upertino

-0.09

 Crunch

-0.09

abeth

-0.09

burgh

-0.09

qry

-0.09

íĥķ

-0.09

arcer

-0.08

osg

-0.08

POSITIVE LOGITS

 values

0.28

 numbers

0.27

 value

0.25

 quantity

0.22

åĢ¼

0.22

values

0.21

 Values

0.19

value

0.19

åĢ¤

0.19

numbers

0.18

Activations Density 0.122%