INDEX

Explanations

references with numbers

The neuron chiefly activates on multi‐digit numerical tokens (e.g., port numbers, timestamps, version or page numbers).

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

-2.86

 also

-2.02

it

-1.79

の方も

-1.70

in

-1.67

 también

-1.52

 help

-1.41

逍遥

-1.38

の方に

-1.34

 периода

-1.31

POSITIVE LOGITS

entown

1.50

鲼

1.45

】：

1.45

ergies

1.44

჻

1.39

鋨

1.38

 بالفعل

1.37

鱚

1.35

♬

1.34

基本的に

1.33

Activations Density 0.074%