INDEX

Explanations

months and dates

The neuron is picking up on standalone numeric tokens in bibliographic metadata—things like volume/issue numbers, years, page numbers, and dates in citations.

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

去年

-1.34

actually

-1.31

饪

-1.30

꜆

-1.29

does

-1.28

↵

-1.26



-1.25

ಱ

-1.23

instead

-1.23

në

-1.22

POSITIVE LOGITS

1.66

as

1.55

on

1.37

At

1.28

 with

1.23

 découvrez

1.18

by

1.16

 апреля

1.13

 October

1.13

 évé

1.13

Activations Density 0.030%