INDEX

Explanations

pronouns followed by verbs

This neuron lights up on unusually rare or uncommon tokens—i.e. words with low overall frequency (high “IDF”) in the text.

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

 when

-2.39

has

-1.73

 while

-1.60

by

-1.54

as

-1.44

if

-1.42

 will

-1.42

at

-1.41

 would

-1.37

 after

-1.35

POSITIVE LOGITS

走吧

1.36

hever

1.29

 necesar

1.27

carus

1.27

ícias

1.22

 ſtill

1.20

 OTHER

1.20

越来越

1.19

 xadrez

1.18

ILERS

1.17

Activations Density 0.028%