INDEX

Explanations

mathematical calculations and code constructs

The neuron detects floating‐point numeric literals (numbers with a decimal point and an “f” suffix) 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

が

-1.45

も

-1.42

all

-1.40

があり

-1.37

商品は

-1.36

調味料

-1.34

などが

-1.33

は

-1.29

がない

-1.24

POSITIVE LOGITS

為に

1.52

 amables

1.45

で見る

1.43

の人に

1.42

 surtido

1.39

んだよね

1.38

んだよ

1.38

卡通

1.36

 عسل

1.35

 delgado

1.33

Activations Density 0.014%