INDEX

Explanations

numbers and citations

This neuron fires on numeric and related spacing tokens—i.e. the volume/issue/year/page numbers (and their thin‐space separators) found in formatted reference entries.

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

ΕΝ

-1.01

Ehrungen

-0.98

uuuuu

-0.96

"_

-0.95

uuuu

-0.92

velocityY

-0.92

jusqu

-0.92

cluye

-0.90

 regeringen

-0.89

 Futsal

-0.89

POSITIVE LOGITS

to

1.08

 even

1.05

esseract

0.92

յու

0.87

同じく

0.86

({});

0.85

资本

0.83

手续

0.81

waż

0.79

 letters

0.78

Activations Density 0.001%