INDEX

Explanations

competition reality shows

The neuron detects capitalized title and heading tokens, particularly proper names and key words in article or section titles.

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

</h3>

-1.41

嚀

-1.28

 chré

-1.20

ությ

-1.20

 zapato

-1.16

{};

-1.15

 córdoba

-1.13

 doká

-1.12

 Decreto

-1.10

趺

-1.10

POSITIVE LOGITS

1.20

while

1.16

ھو

1.10

phytes

1.09

ھر

1.06

karte

1.06

lipse

1.04

注釈

1.03

pesar

1.03

centro

1.03

Activations Density 0.023%