INDEX

Explanations

code structure punctuation

The neuron primarily activates on numeric literals (tokens consisting of digits, including integers and floats).

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

놨

-1.37



-1.35

 bucal

-1.34

 během

-1.29

 langkah

-1.28

 linho

-1.26

癡

-1.26

ㄙ

-1.25

 обладает

-1.24

 cheia

-1.24

POSITIVE LOGITS

to

1.72

at

1.48

or

1.42

Спасылкі

1.39

You

1.38

elif

1.35

 хотя

1.34

勇者

1.34

 persönlichen

1.34

 echten

1.31

Activations Density 0.004%