INDEX

Explanations

enclosed words, or possessives

This neuron fires most strongly on words or short phrases enclosed in quotation marks or apostrophes.

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

</h4>

-1.44

'];

-1.27

</b>

-1.23

</sup>

-1.19

Also

-1.16

%)$

-1.13

eine

-1.13

</em>

-1.07

');

-1.06

ICIÓN

-1.06

POSITIVE LOGITS

起码

1.36

将在

1.27

就没有

1.10

 künftig

1.07

はなく

1.05

 vieler

1.02

 χωρίς

1.02

 hiçbir

1.02

 всех

1.02

对待

1.01

Activations Density 0.101%