INDEX

Explanations

one way to

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

/we

-0.09

ÃĢ

-0.09

 Arist

-0.08

 tras

-0.08

/she

-0.08

igli

-0.08

-0.07

ulen

-0.07

ulus

-0.07

æ²

-0.07

POSITIVE LOGITS

 simple

0.14

 simply

0.13

 arguably

0.13

 understanding

0.13

 ensuring

0.13

 simplement

0.13

 undoubtedly

0.12

 through

0.12

 Simple

0.12

simple

0.11

Activations Density 0.091%