INDEX

Explanations

numbers and years

This neuron detects numeric tokens (e.g., years, percentages, counts, and other number literals).

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

 with

-1.37

年に

-1.25

！！！

-1.23

=\"

-1.20

 mètres

-1.20

-1.18

-1.15

ׇ

-1.15

",

-1.13

:+:

-1.13

POSITIVE LOGITS

 overall

1.92

it

1.79

 total

1.68

 last

1.44

its

1.39

んだから

1.38

 niektó

1.30

がありません

1.23

ύ

1.23

 average

1.15

Activations Density 0.024%