INDEX

Explanations

numbers and identifiers

The neuron primarily fires on long numeric sequences (multi-digit numbers such as timestamps, IDs, or other numeric codes).

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

 that

-1.90

or

-1.76

 descrito

-1.61

 inversores

-1.57

 rozwoju

-1.52

 which

-1.48

-1.47

but

-1.46

 mencionado

-1.42

 concentración

-1.41

POSITIVE LOGITS

 setelah

1.77

这些

1.75

…」

1.65

1.56

 These

1.56

These

1.55

vät

1.51

1.46

два

1.46

 también

1.46

Activations Density 0.038%