INDEX

Explanations

facilitate or connect

The neuron highlights prominent domain‐specific technical keywords—i.e. singular nouns that name the core concepts or "switches" in a passage.

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

 including

-1.29

how

-1.04

 așa

-0.98

 this

-0.97

 massima

-0.93

 another

-0.93

 partea

-0.91

mar

-0.90

有一個

-0.90

using

-0.89

POSITIVE LOGITS

for

1.95

of

1.45

 для

1.40

 bagi

1.30

 både

1.29

вѣ

1.27

 للمعارف

1.27

tothe

1.25

 dorados

1.23

 capable

1.22

Activations Density 0.120%