INDEX

Explanations

code definitions and schema

The neuron activates on numeric literals (especially decimals) in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 other

-2.34

 others

-1.59

 otros

-1.55

 pomys

-1.41

 otras

-1.39

 innych

-1.39

 других

-1.39

沨

-1.39

 veloce

-1.37

 outros

-1.36

POSITIVE LOGITS

though

1.27

 zowel

1.23

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.20

text

1.17

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.16

end

1.14

 either

1.14

</h2>

1.14

↵↵

1.14

↵↵↵↵↵↵↵↵↵↵

1.14

Activations Density 0.026%