INDEX

Explanations

hexadecimal strings, hashes, and identifiers

This neuron detects hexadecimal digit sequences (byte‐pair tokens) as found in checksums, hashes, and other hex‐encoded identifiers.

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

and

-1.88

 while

-1.52

 like

-1.47

為に

-1.47

as

-1.43

for

-1.42

 with

-1.42

to

-1.42

部分を

-1.35

it

-1.34

POSITIVE LOGITS

2.00

 frère

1.76

1.73

 kostenlose

1.65

 verrou

1.65

 âme

1.59



1.59

 formes

1.57

 relâche

1.56

Activations Density 0.012%