INDEX

Explanations

code comments and declarations

The neuron fires on tokens that include digit‐based type or version identifiers—most strongly on fixed‐width integer type names like “uint32_t” and “uint64_t” (and to a lesser extent platform/version macros such as “MSC_VER”).

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

 Also

-1.67

栳

-1.38

𝐈

-1.38

-1.34

is

-1.34

辐

-1.34

都不能

-1.30

was

-1.30

ㄹ

-1.30

 Those

-1.29

POSITIVE LOGITS

 být

1.67

 feitas

1.64

懨

1.62

 месяцев

1.61

試試

1.57

 necessárias

1.55

孥

1.48

欣賞

1.48

 ricette

1.47

 recensioni

1.46

Activations Density 0.001%