INDEX

Explanations

hexadecimal identifiers

The neuron fires on tokens that are hexadecimal sequences (hex digits) as found in GUIDs, hashes, URLs, or other ID strings.

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

as

-1.51

can

-1.48

and

-1.45

 should

-1.42

has

-1.39

 could

-1.38

is

-1.34

may

-1.30

 will

-1.30

 colspan

-1.20

POSITIVE LOGITS

 ícones

1.57

 mejoras

1.55

 about

1.53

also

1.49

ticism

1.48

 deudas

1.47

 marchio

1.45

 clássico

1.44

不僅

1.43

selben

1.41

Activations Density 0.195%