INDEX

Explanations

random generation

The neuron tends to fire strongly on technical, formal or bureaucratic terms—i.e. official/academic vocabulary.

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

ien

0.81

 AudioClip

0.72

 jitter

0.72

 pamię

0.70

że

0.69

 যাতে

0.69

)-

0.69

 যেভাবে

0.67

 paździer

0.67

 fileSize

0.66

POSITIVE LOGITS

ти

0.97

他

0.89

taxes

0.83

banks

0.83

к

0.81

தே

0.80

продук

0.79

হয়

0.79

ઙ

0.78

्यर्थ

0.78

Activations Density 0.000%