INDEX

Explanations

code imports and assertions

The neuron detects standalone multi‐digit integer literals in code.

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

 tragé

-1.32

всем

-1.28

 turist

-1.21

 rése

-1.20

Jeśli

-1.20

мне

-1.18

мл

-1.16

Ι

-1.15

toDouble

-1.15

 biodivers

-1.15

POSITIVE LOGITS

and

1.38

1.20

↵↵

1.17

 magnum

1.07

1.06

 then

1.05

1.01

tr

1.01

↵

1.00

0.98

Activations Density 0.001%