INDEX

Explanations

up to [quantity]

The neuron primarily detects “up to” constructions that introduce a numeric limit or maximum (e.g. “up to 171”).

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

慙

-1.56

 douane

-1.56

 patienter

-1.55

for

-1.46

with

-1.46

basically

-1.40

︕

-1.40

 طريقة

-1.40

during

-1.39

 provinciale

-1.39

POSITIVE LOGITS

 each

1.80

 Each

1.66

 sensaciones

1.60

 there

1.47

 lograr

1.46

 asegurarse

1.42

 území

1.37

 complicaciones

1.36

 considerada

1.36

 further

1.36

Activations Density 0.027%