INDEX

Explanations

end of the

This neuron detects occurrences of the phrase “end of,” i.e. references to the end of some period or segment.

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

by

-1.95

to

-1.84

through

-1.72

and

-1.70

in

-1.69

at

-1.66

no

-1.61

as

-1.61

<i>

-1.52

日まで

-1.49

POSITIVE LOGITS

茕

1.93

 prêts

1.91

 expédi

1.86

 douane

1.84

 achet

1.81

folie

1.79

 Polícia

1.77

孓

1.77

豋

1.75

 désolés

1.73

Activations Density 0.017%