INDEX

Explanations

initialization prefixes

The neuron strongly activates on numeric tokens (digit sequences) in code or markup.

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

ּוֹ

-1.78

ִּ

-1.70

ֶּ

-1.62

 risk

-1.39

ַּ

-1.38

 мечты

-1.34

who

-1.33

かけた

-1.29

 procedures

-1.25

ülerin

-1.23

POSITIVE LOGITS

ׇ

1.36

 chré

1.33

 reflexiones

1.31

 definiciones

1.23

ίδα

1.21

などを

1.18

了下去

1.17

*}\

1.17

着丈

1.16

 сахара

1.15

Activations Density 0.002%