INDEX

Explanations

modular approach

This neuron detects mentions of modular components or sections—words referring to “modules,” “sections,” and similar units.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 morreu

0.82

 benefactor

0.80

 bodyguard

0.78

 धोखा

0.78

 ballerina

0.78

 महिला

0.74

 bomba

0.74

 upheaval

0.74

 foundry

0.73

 மற்றொரு

0.72

POSITIVE LOGITS

それぞれ

1.31

 각각

1.27

それぞれの

1.27

各个

1.26

Each

1.22

 jednotliv

1.22

每个

1.21

 each

1.20

 تقسیم

1.18

each

1.16

Activations Density 1.357%