INDEX

Explanations

parts of words

The neuron activates on numeric tokens conveying quantitative specifications (e.g., measurements, version numbers, capacities).

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

־ה

-1.36

 бампер

-1.36

 зависит

-1.35

 Einige

-1.35

 их

-1.32

 μας

-1.29

 they

-1.29

 fría

-1.27

 Bemerkungen

-1.27

ִּ

-1.27

POSITIVE LOGITS

 your

1.80

you

1.71

 вам

1.45

 vernac

1.38

 pengar

1.35

】:

1.34

 kanak

1.33

 latine

1.32

 יְ

1.30

 Præ

1.29

Activations Density 0.002%