INDEX

Explanations

miniature and small concepts

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

 sore

-0.09

 slam

-0.08

Tur

-0.07

(>

-0.07

 Executive

-0.07

IFI

-0.07

 exec

-0.07

 brill

-0.07

 slug

-0.07

 ditch

-0.07

POSITIVE LOGITS

 disproportionately

0.11

/tiny

0.11

 bons

0.11

 Mini

0.11

(big

0.11

-big

0.10

 miniature

0.10

wes

0.10

.nano

0.10

Nano

0.10

Activations Density 1.159%