INDEX

Explanations

controller operation structure functions

The neuron responds to unusually long, multi‐syllabic tokens (complex words).

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

of

-1.52

for

-1.49

was

-1.41

if

-1.41

приятия

-1.34

 おしゃれ

-1.30

ciem

-1.28

if

-1.19

っぷり

-1.18

Penulis

-1.16

POSITIVE LOGITS

 that

2.05

 your

1.63

 sử

1.54

 such

1.43

 aceea

1.41

 Wartawan

1.38

diğinde

1.33

 utilizar

1.28

 terlaris

1.28

JAMIN

1.27

Activations Density 0.095%