INDEX

Explanations

documentary

The neuron identifies salient content words—especially topic-specific nouns and technical terms that carry the document’s main subject matter.

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

citement

-1.53

氿

-1.52

 others

-1.51

copie

-1.51

᾿

-1.46

 Atención

-1.45

꜆

-1.44

厹

-1.42

</h1>

-1.42

mbgg

-1.41

POSITIVE LOGITS

six

1.84

 time

1.77

not

1.76

for

1.76

でございます

1.74

at

1.70

 every

1.68

 four

1.66

 superbly

1.65

From

1.63

Activations Density 0.089%