INDEX

Explanations

Dona Sarkar, Girls'

The neuron fires on the boilerplate comment‐header and metadata tokens (license text, author tags, copyright years, uppercase keywords) in code.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

했

-0.77

主人公

-0.76

 thầy

-0.75

 thải

-0.72

 Junge

-0.71

noinspection

-0.69

obar

-0.69

 penyanyi

-0.69

 جدی

-0.68

 Bersama

-0.68

POSITIVE LOGITS

 Enrique

0.81

tiennent

0.80

kuuta

0.80

WPF

0.79

 Token

0.78

Dp

0.77

phu

0.77

))=

0.77

 RUSSELL

0.75

 مم

0.74

Activations Density 0.018%