INDEX

Explanations

common words & difficult concepts

The neuron fires strongly on promotional or evaluative adjectives and adverbs (e.g. “free,” “new,” “exciting,” “easy,” “tuyệt/vui vẻ”) used to boost marketing appeal.

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

 skirt

0.61

 aggravated

0.59

 lizard

0.58

 ridge

0.57

 polypeptide

0.57

 picket

0.57

 cartridge

0.56

 patiently

0.56

 allegiance

0.56

 knowingly

0.56

POSITIVE LOGITS

 banget

1.40

な

1.40

无比

1.37

ที่สุด

1.27

 ترین

1.27

ترین

1.20

 ביותר

1.19

‌ترین

1.17

 للغاية

1.14

มากๆ

1.13

Activations Density 0.076%