INDEX

Explanations

refers to

The neuron fires strongly on high-level “buzz” nouns (like luxury, artistry, performance, goal, effect, etc.) that label features or benefits in marketing or promotional copy.

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

必备

0.77

ໄລ

0.76

 चौहान

0.71

 শরণ

0.70

 ມີ

0.70

 अधिकारी

0.69

 آفیسر

0.69

 コーヒー

0.68

 הי

0.66

 ssize

0.66

POSITIVE LOGITS

keras

0.86

 engen

0.80

schrift

0.77

 kamp

0.75

aligned

0.75

sche

0.75

keit

0.74

 keresztül

0.73

mechan

0.73

schaft

0.72

Activations Density 0.000%