INDEX

Explanations

The neuron detects formal mathematical terminology—verbs and nouns like “satisfies,” “assumption,” “equation,” “recurrence,” and “inequality” that mark technical statements.

satisfies conditions and constraints

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

𨭎

-1.47

 !!!!!

-1.37

嬲

-1.33

嗑

-1.25

逃跑

-1.22

 usando

-1.21

 dando

-1.20

ciating

-1.16

 puramente

-1.16

uristic

-1.15

POSITIVE LOGITS

the

1.84

 that

1.52

Being

1.41

 with

1.41

 both

1.40

𓄹

1.37

 myös

1.35

 Therapie

1.32

袮

1.32

Activations Density 0.012%