INDEX

Explanations

newbie progress

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 gossip

-0.12

 behind

-0.12

 grooming

-0.11

 Behind

-0.11

 Cute

-0.11

illac

-0.11

 mutual

-0.11

 Mutual

-0.10

 discreet

-0.10

 pawn

-0.10

POSITIVE LOGITS

 newbie

0.17

 beginners

0.14

 progress

0.14

 plateau

0.14

 taper

0.13

 regimen

0.13

 experimentation

0.13

 beginner

0.13

 experimenting

0.13

 posted

0.13

Activations Density 0.877%

No Known Activations