INDEX

Explanations

discourse markers and descriptions

This neuron detects vague qualifiers like “sort of” and “kind of.”

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

He

-1.72

-1.70

</h3>

-1.60

-1.56

-1.38

te

-1.38

 this

-1.31

了他

-1.31

-1.30

的時候

-1.28

POSITIVE LOGITS

𖣘

1.81

%%%%%%%%%

1.63

cardigan

1.59

to

1.56

 ungkap

1.55

mitsubishi

1.52

viking

1.51

⎆

1.48

Français

1.44

%%%%%%%

1.42

Activations Density 0.012%