INDEX

Explanations

as possible

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

-0.09

ensis

-0.09

 Biggest

-0.09

aten

-0.08

 Benedict

-0.08

lette

-0.08

.serialize

-0.08

 Wilkinson

-0.08

alle

-0.08

 barely

-0.07

POSITIVE LOGITS

 gonna

0.23

gon

0.16

 destined

0.14

 going

0.13

 ever

0.13

 EVER

0.13

-going

0.13

Going

0.12

ductive

0.11

 doomed

0.11

Activations Density 0.011%