INDEX

Explanations

headings and introductory phrases

The neuron is triggered by the “0x” prefix of hexadecimal numeric literals in code.

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

↵

-1.90

</h2>

-1.47

 asserted

-1.40

انا

-1.38

哋

-1.36

ੲ

-1.33

 deals

-1.32

 nasze

-1.31

tirer

-1.30

 opponent

-1.28

POSITIVE LOGITS

all

1.49

are

1.38

newblock

1.34

</th>

1.31

 لدي

1.27

蒟

1.20

 menaces

1.19

すぎて

1.16

Twas

1.13

 doctrina

1.13

Activations Density 0.001%