INDEX

Explanations

hindsight perspective retrospective

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

jp

-0.09

rah

-0.09

ski

-0.08

/modal

-0.08

 unheard

-0.08

 enclosure

-0.08

/internal

-0.08

 Horton

-0.08

remain

-0.08

 edible

-0.07

POSITIVE LOGITS

 hindsight

0.45

 retrospect

0.39

 retrospective

0.34

 retros

0.30

indsight

0.27

 perspective

0.22

 retro

0.20

 Perspective

0.19

 perspectives

0.17

 distance

0.17

Activations Density 0.147%