INDEX

Explanations

numbers and dates

The neuron fires strongly on proper names, titles, and other salient content words (e.g. names of people, creative works, places) that typically appear in headlines or lead‐in phrases.

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

水墨

-1.13

$};

-1.08

 मुख्य

-1.08

字幕

-1.04

 कि

-1.01

ieť

-1.00

">

-0.98

ји

-0.98

кою

-0.98

 شما

-0.96

POSITIVE LOGITS

 everything

1.31

 hiçbir

1.27

 cristianismo

1.23

 every

1.22

 semillas

1.08

 vegetales

1.05

 исключительно

1.05

྄

1.04

 chrétien

1.04

зий

1.03

Activations Density 0.149%