INDEX

Explanations

documentation markers like `@` and `*/`

The neuron fires on documentation‐style markup tokens—things like asterisks in comment blocks, “@” annotation markers, section headings (e.g. Arguments), and other docstring punctuation.

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

 höl

-1.23

now

-1.22

on

-1.08

Sábado

-1.05

 first

-1.03

 plötzlich

-1.02

最近は

-1.02

UCION

-1.02

 good

-1.00

 because

-0.98

POSITIVE LOGITS

{@

1.17

的照片

1.15

的实力

1.12

 néanmoins

1.11

ありません

1.09

随即

1.09

 accompag

1.08

的气息

1.07

kauft

1.06

 Ejemplo

1.05

Activations Density 0.011%