INDEX

Explanations

This neuron detects numeric tokens (numbers, years, page or figure references, etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 latter

0.50

なのです

0.39

后者

0.38

 válido

0.37

 validInput

0.37

 ಅಥವಾ

0.36

න්නේ

0.36

 assim

0.36

த்தின்

0.35

过的

0.35

POSITIVE LOGITS

.,

1.65

.;

1.36

.:

1.31

.,"

1.27

.),

1.23

.},

1.08

.)

1.07

.).

1.07

./

1.05

.-

1.05

Activations Density 1.101%