INDEX

Explanations

shell commands and scripts

The neuron fires on command‐line usage patterns—things like “./script.sh”, square brackets, version placeholders, and other usage/help‐style tokens.

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

 baratas

-1.33

这里是

-1.27

 вашего

-1.24

 всегда

-1.24

 possibilidade

-1.23

 possibly

-1.23

После

-1.18

 новых

-1.17

 excessively

-1.16

 моего

-1.16

POSITIVE LOGITS

to

1.88

'',

1.51

"",

1.42

 einerseits

1.33

ціона

1.30

 Veränder

1.28

lassified

1.27

Външни

1.27

[],

1.27

 pubblicato

1.27

Activations Density 0.086%