INDEX

Explanations

listing occurrences

The neuron fires strongly on the prepositions, adverbs, and linking words that introduce situational or contextual phrases—especially those marking place, time, or extent (e.g. “filled to,” “places like,” “up to,” “this year,” “in right now”).

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

who

-1.55

 only

-1.46

 will

-1.45

 same

-1.39

-1.33

</h4>

-1.32

</b>

-1.23

 didn

-1.23

</sup>

-1.23

 только

-1.23

POSITIVE LOGITS

 these

2.27

 each

1.84

 this

1.77

 those

1.74

()

1.55

 heilige

1.52

 such

1.51

=>

1.46

 beker

1.45

 Belgische

1.41

Activations Density 0.126%