INDEX

Explanations

names and terms of endearment followed by punctuation

The neuron fires on capitalized proper names or named entities.

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

 ebenfalls

0.56

কারী

0.54

กัน

0.51

 देतात

0.51

ತನ

0.50

 செய்தனர்

0.50

েরই

0.50

 zichzelf

0.50

asInstanceOf

0.49

щают

0.49

POSITIVE LOGITS

ơi

1.48

,</

1.26

 please

1.18

 તમે

1.16

 hãy

1.09

‌,

1.08

 jangan

1.07

please

1.06

，

1.05

Activations Density 0.124%