INDEX

Explanations

magnitude

The neuron strongly fires on multi‐digit numeric tokens (e.g. years, measurements, or other long sequences of digits).

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

assertEqual

-0.77

ствуют

-0.66

ствовали

-0.66

Ґ

-0.66

 Projection

-0.66

गर

-0.65

क्ष

-0.65

 lengthen

-0.65

 Tach

-0.64

 beraten

-0.64

POSITIVE LOGITS

 magnitude

4.66

magnitude

3.67

 Magnitude

3.59

 magnitudes

3.45

Magnitude

3.38

 magnitud

2.50

mag

2.39

mag

2.25

Mag

1.99

MAG

1.95

Activations Density 0.063%