INDEX

Explanations

numbers and code generation

The neuron fires on numeric literals and related code‐discussion tokens (e.g. loop counters, initializations, increments).

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

$=$

0.59

 sparsely

0.57

 innumerable

0.56

 nomes

0.56

 baseUrl

0.55

[`

0.54

$[\

0.54

 तह

0.54

 negated

0.54

 Deleg

0.53

POSITIVE LOGITS

۵

1.46

۶

1.40

۷

1.38

۸

1.37

 ۱۵

1.34

۴

1.33

۹

1.31

 ۱۹

1.25

۳

1.24

७

1.23

Activations Density 0.186%