INDEX

Explanations

program language constructs

The neuron activates on tokens that are names of abstract syntax tree node types or language construct labels (e.g. ArrayExpression, BlockStatement, arrowFunctionBody).

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

who

-1.63

 lino

-1.45

Eerste

-1.41

 paliwa

-1.36

joyed

-1.32



-1.32

 які

-1.29

躥

-1.27

會有

-1.27

 abstracto

-1.26

POSITIVE LOGITS

to

1.57

an

1.57

 last

1.53

or

1.52

in

1.43

the

1.41

 Using

1.41

味付け

1.40

1.35

Activations Density 0.034%