INDEX

Explanations

java package names

The neuron fires on the two‐letter top‐level package identifiers (e.g. “co.” or “cn.”) in Java import/package lines.

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

for

-2.33

 because

-2.22

but

-2.13

 there

-1.90

 first

-1.82

 after

-1.81

as

-1.72

There

-1.68

-1.67

one

-1.67

POSITIVE LOGITS

the

1.86

 größe

1.81

鲧

1.73

uſ

1.64

 diverses

1.61

 frança

1.59

 erhi

1.59

 uſe

1.57

 küche

1.55

every

1.53

Activations Density 0.026%