INDEX

Explanations

historical entities and company names

The neuron fires most strongly on content words that appear as part of titles or headings (e.g. book‐ and poem‐titles, section headings, advertisement headlines).

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

iſter

-1.59

 abbraccio

-1.48

 zijne

-1.47

 that

-1.39

ꦗ

-1.38

罴

-1.36

 podríamos

-1.30

cargas

-1.28

sistente

-1.27

JULY

-1.25

POSITIVE LOGITS

you

1.66

al

1.45

𝓭

1.13

章节

1.13

1.11

 каждого

1.10

cayaan

1.08

【

1.06

one

1.05

 вас

1.05

Activations Density 0.055%