INDEX

Explanations

Beginning phrases or specific titles

The neuron strongly activates on left‐curly‐brace tokens (e.g. “{”), especially in contexts like LaTeX math expressions.

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

and

-1.62

-1.41

 When

-1.41

 Usually

-1.33

if

-1.28

 because

-1.27

 used

-1.24

 Sometimes

-1.23

 Often

-1.20

Some

-1.18

POSITIVE LOGITS

 (“

1.47

 három

1.40

 trzech

1.38

 polskich

1.37

—“

1.35

 ‘’

1.35

including

1.34

 nieuwe

1.34

 tredje

1.34



1.31

Activations Density 0.000%