INDEX

Explanations

the digit 3

This neuron detects numbered labels or headings—tokens representing digits in structural markers like “Part 3,” “Type 2,” table numbers, section IDs, RDD indexes, etc.

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

-2.55

 starker

-1.55

 each

-1.52

けれ

-1.49

Пита

-1.42

Джерело

-1.42

来週

-1.39

Размер

-1.38

 最後

-1.35

Марина

-1.34

POSITIVE LOGITS

lahoma

1.56

 kabát

1.54

you

1.46

ctory

1.45

1.43

ribute

1.41

 ofta

1.41

is

1.40

 acabado

1.40

ㅁ

1.38

Activations Density 0.047%