INDEX

Explanations

std imports

This neuron detects large runs of leading whitespace (i.e. long indentation spans).

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

</h2>

-1.66

 there

-1.55

or

-1.51

ных

-1.46

on

-1.40

 pinnacle

-1.38

<bos>

-1.31

</b>

-1.27

 even

-1.27

乃至

-1.25

POSITIVE LOGITS

 родителей

1.42

ज़

1.40

Twenty

1.39

==>

1.38

Everybody

1.38

 oben

1.34

two

1.32

 zealand

1.30

 letech

1.30

Four

1.30

Activations Density 0.070%