INDEX

Explanations

formal actions and agreements

The neuron consistently picks out the main-event or topic nouns in news headlines and lead sentences (e.g. lawsuit, bug, initiative, pact, strategy, etc.).

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

and

-1.55

as

-1.49

or

-1.47

-1.38

 like

-1.36

-1.34

 what

-1.30

-1.25

-1.24

POSITIVE LOGITS

 that

1.91

 sabun

1.49

ilding

1.43

 effici

1.40

uli

1.39

 arru

1.38

 capilar

1.37

 pourtant

1.37

 konfigura

1.34

romatic

1.33

Activations Density 0.075%