INDEX

Explanations

numbers followed by units or abbreviations

The neuron primarily fires on numeric tokens (e.g. integer and decimal measurements) in the text.

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

МЕН

-1.22

 nomine

-1.19

 encountering

-1.16

了一口

-1.15

UNICIP

-1.13

プール

-1.13

Someone

-1.13

mathrm

-1.12

</td>

-1.11

etc

-1.11

POSITIVE LOGITS

 bicara

1.43

 tertawa

1.27

archical

1.23

Вариан

1.23

at

1.21

 bertanya

1.19

 jät

1.16

 menangis

1.13

ökning

1.10

 parha

1.10

Activations Density 0.131%