INDEX

Explanations

numbers and specific meanings

The neuron fires on the kind of high‐level, definitional or overview phrases that introduce or explain a topic (e.g. “In the … series. This story arc illustrates …,” “and contains …,” “to reinforce …,” “have been part …”).

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

any

-1.97

 this

-1.88

 what

-1.82

 only

-1.64

our

-1.59

 where

-1.55

 when

-1.55

 всі

-1.52

 some

-1.46

 just

-1.45

POSITIVE LOGITS

Beschrijving

1.55

 koruyucu

1.46

 сказать

1.42

 gyvento

1.42

stånd

1.38

 valsty

1.37

ressive

1.35

was

1.35

 зона

1.34

 enlever

1.34

Activations Density 0.003%