INDEX

Explanations

"word" followed by "

The neuron tends to fire on the key topic noun of a passage—the single word that names its main subject (e.g. “ticket,” “candidate,” “skill,” “system,” “color”).

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

aka

0.64

AKA

0.64

etc

0.64

da

0.62

im

0.60

defined

0.59

to

0.58

ILLE

0.58

okens

0.58

uddin

0.57

POSITIVE LOGITS

结构

0.64

संख्या

0.64

 обстоя

0.60

 hierarchy

0.59

数は

0.59

令和

0.59

쉽

0.59

folge

0.57

 upravo

0.56

 structure

0.55

Activations Density 0.423%