INDEX

Explanations

structured data with id and name

The neuron fires on the “{” tokens that open object literals (in JSON or JavaScript).

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

↵

-2.05

硖

-1.58

↑↑↑</

-1.53

gemeen

-1.49

蹉

-1.48

 слегка

-1.48

бую

-1.48

ственного

-1.43

बस

-1.43

CurtirCurtir

-1.41

POSITIVE LOGITS

 with

1.86

</h2>

1.59

1.58

at

1.58

ing

1.49

ଃ

1.48

ferences

1.45

 that

1.45

entrums

1.44

</h3>

1.42

Activations Density 0.008%