INDEX

Explanations

code or questions

The neuron activates strongly on code-style tokens—comment markers, punctuation (like “.” and “;”), braces, and common programming keywords—i.e. boilerplate/source-code syntax.

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

ֺ

-1.52

it

-1.39

 τὴν

-1.38

🧌

-1.28

 their

-1.28

🫴

-1.28

 всі

-1.27

 δὲ

-1.27

และ

-1.24

 τὸν

-1.24

POSITIVE LOGITS

 сторона

1.52

 после

1.48

meanor

1.40

 nampak

1.39

 иногда

1.38

после

1.31

что

1.30

 стакан

1.30

 неприят

1.28

によって

1.26

Activations Density 0.011%