INDEX

Explanations

not aware

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 refrain

-0.12

 GENERIC

-0.11

bil

-0.10

 grac

-0.10

 egal

-0.09

 obscured

-0.09

 unimagin

-0.09

generic

-0.09

isen

-0.09

 unin

-0.09

POSITIVE LOGITS

 notice

0.15

 Notice

0.14

 noticing

0.13

 knowledge

0.13

notice

0.13

Notice

0.12

 obvious

0.12

Knowledge

0.12

 notices

0.12

 noticed

0.11

Activations Density 0.020%