INDEX

Explanations

questions about oneself

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

bl

-0.09

 sensit

-0.09

 distr

-0.09

 Wise

-0.09

 sensitivity

-0.09

 Sweat

-0.09

 Sheep

-0.09

ãĥ»ãĥ»ãĥ»\n\n

-0.09

ux

-0.09

 Erot

-0.08

POSITIVE LOGITS

 boredom

0.20

 bored

0.19

toy

0.16

 toys

0.16

 interest

0.16

toy

0.16

 enjoy

0.15

 amusement

0.14

 enjoying

0.13

 play

0.13

Activations Density 0.074%