INDEX

Explanations

with + descriptive phrase

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

md

-0.09

Lob

-0.08

sip

-0.08

 thrust

-0.08

oland

-0.07

anned

-0.07

cel

-0.07

ESP

-0.07

 chap

-0.07

 Cable

-0.07

POSITIVE LOGITS

 everywhere

0.12

-relative

0.12

 Lots

0.11

 hundreds

0.11

 plenty

0.11

 lots

0.11

oret

0.10

rant

0.10

Lots

0.10

hus

0.10

Activations Density 0.001%

No Known Activations