INDEX

Explanations

The neuron flags very common, high‐frequency function words (particularly short articles, prepositions, and similar stopwords).

a word followed by another

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

Ausstattung

-1.45

Meskipun

-1.38

Fakta

-1.34

 pérd

-1.32

Selain

-1.30

 évêque

-1.28

こういった

-1.27

 薬局

-1.26

が上が

-1.25

 間取り

-1.24

POSITIVE LOGITS

1.21

1.18

to

1.16

1.08

And

1.05

𝚒

1.04

1.03

Activations Density 0.075%