INDEX

Explanations

inventory, logistics, and operational terms

The neuron selectively fires on programming‐style identifiers—like table or column names, class/function names, URL segments, and other code tokens.

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

 étant

-1.37

 έχουν

-1.33

 chacun

-1.33

gggg

-1.28

ländischen

-1.26

∎

-1.25

bj

-1.24

🤯

-1.24

ㅋㅋ

-1.23

🌞

-1.23

POSITIVE LOGITS

of

1.70

on

1.44

for

1.44

 from

1.39

 Ausblick

1.36

 Hinblick

1.32

！！！”

1.29

 vermeiden

1.27

 unterstützen

1.26

 Fälle

1.26

Activations Density 0.008%