INDEX

Explanations

specific word beginnings

The neuron lights up on low‐frequency, content‐heavy tokens—rare words, names, numbers, and other distinctive or technical terms.

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

for

-1.91

 without

-1.69

 with

-1.59

byes

-1.45

 before

-1.44

-1.43

by

-1.39

 through

-1.35

ità

-1.25

 untuk

-1.24

POSITIVE LOGITS

 ozdob

1.29

 сообщил

1.29

 fidé

1.28

kac

1.26

 camuf

1.26

}/*

1.20

jds

1.20

Lastly

1.20

来歴

1.20

 tampaknya

1.19

Activations Density 0.326%