INDEX

Explanations

dates and programming comments

The neuron activates on numeric literals (tokens consisting of digits, often including decimal points).

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

to

-2.00

̐

-1.40

by

-1.34

 området

-1.34

</h1>

-1.34

(/^

-1.29

 based

-1.28

 then

-1.27

 esso

-1.27

if

-1.25

POSITIVE LOGITS

1.34

MOST

1.30

'])

1.27

伱

1.27

 modifikasi

1.25

 megfelelő

1.25

 أيض

1.25

：

1.24

piecze

1.23

 Stoffe

1.23

Activations Density 0.010%