INDEX

Explanations

names starting with P, M, D

This neuron activates on numeric tokens—particularly year and date numbers.

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

 senin

-1.40

 vulga

-1.39

 klinik

-1.30

 diskusi

-1.29

 abiye

-1.25

 etek

-1.25

 praktis

-1.25

 conducteur

-1.24

 rése

-1.23

OTHY

-1.22

POSITIVE LOGITS

but

1.65

1.32

It

1.29

 Even

1.28

1.26

as

1.25

Ar

1.24

1.23

Bar

1.23

1.22

Activations Density 0.315%