INDEX

Explanations

both

This neuron detects numeric tokens and digit sequences (e.g. version numbers, measurements, counts).

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

to

-1.58

by

-1.38

 with

-1.23

all

-1.18

 when

-1.11

so

-1.10

 that

-1.06

at

-1.05

in

-1.00

as

-0.99

POSITIVE LOGITS

BOTH

1.31

両方

1.30

both

1.24

gruppen

1.24

 BOTH

1.19

 begge

1.14

 både

1.13

 ambos

1.12

 entrambi

1.12

 beide

1.10

Activations Density 0.062%