INDEX

Explanations

recognizing typeof

The neuron activates on occurrences of the “typeof(…)” operator 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

 literalmente

-1.79

{*}

-1.69

来自于

-1.63

 preparando

-1.58

 distinto

-1.58

Fakta

-1.52

 zprávy

-1.52

værende

-1.50

 verksamhet

-1.50

同上

-1.50

POSITIVE LOGITS

of

2.20

和

2.00

 possible

1.64

、

1.57

ા

1.52

的

1.51

ल

1.46

you

1.45

</h1>

1.45

the

1.43

Activations Density 0.001%