INDEX

Explanations

return statements and computations

The neuron activates on numeric literals (integer or floating-point constants) 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

)};

-1.58

sistence

-1.45

fatalError

-1.44

黩

-1.43

település

-1.41

 dieser

-1.41

itename

-1.37

ntax

-1.34

וּ

-1.30

///<

-1.30

POSITIVE LOGITS

 then

1.92

 also

1.60

 zrobić

1.47

alucía

1.45

oczes

1.41

entown

1.39

 chiede

1.39

thodox

1.38

当时

1.34

べて

1.32

Activations Density 0.039%