INDEX

Explanations

look for information or evidence

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 ters

-0.11

uen

-0.11

 Circ

-0.10

Wo

-0.10

 vivid

-0.09

 excellent

-0.09

 shortly

-0.09

 Sunny

-0.09

Gs

-0.09

 mono

-0.09

POSITIVE LOGITS

 somehow

0.16

 clues

0.15

 signs

0.15

æĺ¯åĲ¦

0.14

any

0.14

 æĺ¯åĲ¦

0.13

 indication

0.12

 whether

0.12

 expecting

0.11

 maybe

0.11

Activations Density 0.059%

No Known Activations