INDEX

Explanations

code snippets

The neuron is picking up on leading indentation (i.e. runs of whitespace at the start of code lines).

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

 this

-1.86

 about

-1.51

or

-1.48

 these

-1.35

🔥🔥

-1.30

嘭

-1.27

lysm

-1.27

ských

-1.26

神色

-1.23

</em>

-1.23

POSITIVE LOGITS

.}(

1.95

MAXN

1.88

नलोड

1.70

鈊

1.59

toDouble

1.54

諼

1.50

႐

1.48

tgärder

1.48

$};

1.46

🢒

1.46

Activations Density 0.131%