INDEX

Explanations

such as and like

The neuron fires on constructions where two named entities or items are paired together—i.e. “X and Y” or “X & Y” coordination.

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

OLIC

-1.79

źdz

-1.45

 wasnt

-1.44

ίζει

-1.41

 stupidity

-1.40

俵

-1.39

 accessoire

-1.39

(*´

-1.38

遽

-1.38

 versión

-1.37

POSITIVE LOGITS

are

2.58

 that

2.42

to

2.00

 innymi

1.90

 which

1.86

our

1.73

 such

1.65

 already

1.58

 others

1.50

of

1.45

Activations Density 0.073%