INDEX

Explanations

numbers after col

The neuron fires on numeric literals consisting of three or more digits.

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

how

-1.70

 after

-1.30

 znaleźć

-1.30

as

-1.20

 there

-1.13

 like

-1.12

 fifteen

-1.10

 pamię

-1.10

 every

-1.08

 rijden

-1.07

POSITIVE LOGITS

2.17

۲

1.69

 två

1.48

²,

1.40

２

1.28

 visse

1.27

²

1.27

 dioxide

1.27

 poë

1.22

².

1.21

Activations Density 0.162%