INDEX

Explanations

prints and code execution

The neuron fires on runs of leading or in‐line whitespace (indentation) 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

 such

-1.45

of

-1.41

 Some

-1.38

UnknownFieldSet

-1.29

};

-1.28

 from

-1.27

 по

-1.27

$.

-1.26

);

-1.25

フェクト

-1.23

POSITIVE LOGITS

鹀

1.44

⋗

1.42

 verir

1.38

 verrou

1.34

derived

1.33

 puedo

1.32

тивные

1.31

 puedes

1.31

 quiere

1.30

 tamaño

1.30

Activations Density 0.027%