INDEX

Explanations

depends on factors

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

trap

-0.12

 epic

-0.10

opi

-0.10

 extremes

-0.10

Trap

-0.09

 trap

-0.09

 wedge

-0.09

 fairly

-0.08

 ãĢĢ

-0.08

 onFocus

-0.08

POSITIVE LOGITS

 variables

0.21

 Variables

0.18

variables

0.17

Variables

0.17

 future

0.14

 unknow

0.14

 uncont

0.13

 factors

0.13

 fluid

0.13

 unknown

0.12

Activations Density 0.081%