INDEX

Explanations

annual meeting

The neuron strongly activates on salient nouns—especially technical terms or proper‐name nouns (e.g. “Meeting,” “Machine,” “macrophage,” “life table,” “special purpose,” “uptake,” “punt”).

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

 thus

-1.46

 however

-1.46

 therefore

-1.37

 largely

-1.36

to

-1.36

 esche

-1.30

particularly

-1.28

 about

-1.28

 notamment

-1.27

 somewhat

-1.27

POSITIVE LOGITS

Both

1.80

 with

1.75

 Both

1.73

 through

1.67

 medlemmer

1.59

 cykel

1.59

 priser

1.56

 sociala

1.54

 både

1.53

 HUGE

1.49

Activations Density 0.100%