INDEX

Explanations

fellow members of a group

The neuron detects references to people as part of a group—especially the word “fellow” followed by a social category (e.g. fellow humans, fellow Christians, fellow teens).

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

 tevreden

-1.48

anyaan

-1.40

虯

-1.39

ToProps

-1.32

 automáticamente

-1.29

 suoi

-1.29



-1.29

ventude

-1.28

 اسلامی

-1.27

்க

-1.27

POSITIVE LOGITS

 contou

1.63

 protesta

1.35

 russo

1.31

 fellow

1.30

 pretende

1.30

ADRID

1.30

 lends

1.29

1.27

 disgruntled

1.27

 colleagues

1.25

Activations Density 0.011%