INDEX

Explanations

This neuron primarily responds to evaluative adjectives and adverbs—that is, sentiment-laden descriptive words conveying opinion or emphasis.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 einfachen

1.49

 celulares

1.40

 integrantes

1.38

 officielles

1.34

 posibles

1.34

 baddies

1.32

 grandes

1.29

 mesmos

1.26

 civiles

1.26

 importantes

1.25

POSITIVE LOGITS

ly

1.95

amente

1.59

est

1.58

‌ترین

1.55

ity

1.40

ভাবে

1.35

 ترین

1.35

な

1.35

ที่สุด

1.31

पणे

1.30

Activations Density 14.407%