INDEX

Explanations

letter sequences and symbols

The neuron fires specifically on regex metasyntax symbols—especially the brackets, carets, backslashes, and related tokens that form character-class and other regex operators.

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

 parceria

0.65

 오는

0.64

 rebord

0.64

 situazione

0.61

 adoration

0.61

ರುವುದು

0.61

 फेक

0.61

 어머니

0.59

 previsto

0.59

 disfrutar

0.58

POSITIVE LOGITS

ABCDEFGHIJKLMNOP

1.24

abcdef

1.13

ABC

0.96

ABCD

0.89

xyz

0.88

ABC

0.84

ijkl

0.81

XYZ

0.81

abcd

0.78

uvwxyz

0.76

Activations Density 0.016%