INDEX

Explanations

mathematical expressions with quantifiers

The neuron is detecting LaTeX math‐mode tokens—commands (like \forall, \quad, \tag) and alignment/symbol markers (&, braces, etc.) in formulas.

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

 \%}$

-1.58

 charmant

-1.44

❔

-1.38

စ်

-1.37

 Nedir

-1.35

 bluza

-1.34

빕

-1.31

 БА

-1.30

 toit

-1.29

<{

-1.29

POSITIVE LOGITS

1.65

 because

1.61

or

1.58

 from

1.47

to

1.46

now

1.46

1.43

see

1.39

why

1.37

 saying

1.34

Activations Density 0.071%