INDEX

Explanations

what followed by clarification

This neuron detects uses of “what” introducing a descriptive or renaming clause (e.g. “what later was called…,” “what is now…”).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

Grazie

0.89

た

0.87

Thanks

0.86

에

0.84

い

0.84

ಈ

0.77

ecause

0.77

Yay

0.77

ぅ

0.77

𝘴

0.77

POSITIVE LOGITS

 constit

0.89

 infrequently

0.72

 Steinberg

0.71

 primarily

0.71

 termed

0.70

 Primarily

0.70

 purported

0.68

 INVALID

0.67

 infrequent

0.67

 Stub

0.66

Activations Density 0.074%