INDEX

Explanations

but, it is

The neuron is broadly detecting common English function words—especially auxiliary and connective tokens like “is,” “to,” “would,” “if,” “it,” “but,” and similar.

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

but

-3.03

 however

-2.59

 though

-2.11

ですが

-2.05

But

-2.03

 infatti

-2.02

 però

-2.00

 ولكن

-1.95

 However

-1.86

 azonban

-1.77

POSITIVE LOGITS

 will

1.68

are

1.59

it

1.59

terscotch

1.44

lorin

1.38

its

1.34

actéristi

1.34

 bloemen

1.33

 donker

1.32

érée

1.30

Activations Density 1.333%