INDEX

Explanations

what can be done

The neuron detects modal and semi-modal auxiliary words (e.g. can, should, would, have to, like to) expressing ability, permission, or obligation.

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

-1.97

 gesamte

-1.70

逦

-1.69

萏

-1.62



-1.62

芣

-1.55

ньому

-1.52

猕

-1.50

ও

-1.48

 onun

-1.48

POSITIVE LOGITS

 which

1.93

has

1.79

and

1.72

 shows

1.52

 uses

1.51

 represents

1.49

Now

1.48

𝙧

1.47

 welchen

1.47

 whose

1.47

Activations Density 0.031%