INDEX

Explanations

core values and principles

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

 Prel

-0.11

 proc

-0.10

?type

-0.10

 greatness

-0.09

wor

-0.09

 Intro

-0.09

(END

-0.09

 prelim

-0.09

oph

-0.09

Intro

-0.09

POSITIVE LOGITS

 acronym

0.14

 principle

0.11

 harm

0.11

 principles

0.11

 Consent

0.11

 sane

0.11

 Harm

0.11

 harmless

0.11

 triplet

0.11

 mantra

0.11

Activations Density 0.036%