INDEX

Explanations

change from different

This neuron detects phrases that signal a change, contrast, or novelty (e.g. “change of pace,” “something a little bit different,” “break from …”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 where

-1.48

-1.27

 jopa

-1.24

 that

-1.24

you

-1.19

 following

-1.15

—

-1.09

ers

-1.08

 could

-1.06

--

-1.05

POSITIVE LOGITS

 zbyt

1.28

 Niveau

1.26

GOTREF

1.20

Deux

1.20

面目

1.18

 than

1.18

 passend

1.18

wechs

1.17

lables

1.16

zyński

1.16

Activations Density 0.043%