INDEX

Explanations

variable names and data structures

The neuron is picking out code tokens that consist of or begin with uppercase letters—e.g. type/class/variable names in PascalCase or ALL-CAP identifiers.

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

 duża

-1.35

jualan

-1.30

 delikat

-1.28

ところに

-1.25

為に

-1.24

 resultaat

-1.23

fibra

-1.23

or

-1.20

-1.18

 zimowa

-1.17

POSITIVE LOGITS

 oktober

1.34

 tribu

1.34

 aras

1.26

 ↓,

1.25

 Coruña

1.17

 anormal

1.16

 tragedia

1.16

 slår

1.16

さぎ

1.15

vska

1.15

Activations Density 0.037%