INDEX

Explanations

convenience

The neuron fires on English words with common “-ing,” “-ness,” and “-ence” suffixes that mark gerunds or abstract nouns.

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

 following

-1.45

having

-1.34

 most

-1.29

 sommige

-1.27

など

-1.27

 disponibile

-1.25

at

-1.24

 last

-1.20

thanks

-1.19

allowing

-1.18

POSITIVE LOGITS

 those

1.63

𝖋

1.50

 nervioso

1.40

 dibujado

1.38

Président

1.38

日目

1.37

řízen

1.37

关于

1.37

 např

1.36

lccc

1.32

Activations Density 0.159%