INDEX

Explanations

for years and weeks

This neuron activates on introductory temporal-duration phrases (e.g. “For a long time,” “For many years,” 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

edoria

-1.54

書いた

-1.45

 from

-1.43

one

-1.42

رها

-1.34

鲥

-1.34

 preocupado

-1.30

Ʊ

-1.30

 architect

-1.30

eniging

-1.26

POSITIVE LOGITS

no

1.70

懨

1.50

 jusqu

1.46

ずっと

1.46

Instead

1.41

eacute

1.41

 instead

1.40

using

1.34

という

1.33

};

1.32

Activations Density 0.041%