INDEX

Explanations

that is

The neuron fires on tokens of the phrase “that is” (and its equivalents like Chinese “那就”), i.e. on the discourse marker introducing a clarification or restatement.

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

^{\

-1.72

of

-1.63

-1.62

atürk

-1.59

霂

-1.57

ició

-1.55

 would

-1.52

 }^{\

-1.48

 }^{*}

-1.48

 stava

-1.47

POSITIVE LOGITS

 Estas

1.43

 passend

1.41

まるで

1.41

1.36

 echte

1.34

 technisch

1.31

 enliven

1.30

 mieście

1.29

まじ

1.28

 gewisser

1.28

Activations Density 0.001%