INDEX

Explanations

Q: ` or WARNING

The neuron detects numeric literals and numeric sequences (including hexadecimal and decimal constants) in code.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

這

-1.48

</h4>

-1.45

皤

-1.37

嬲

-1.34

เกง

-1.27

 بعض

-1.25

儈

-1.24

isierte

-1.23

 Direktur

-1.23

͞

-1.21

POSITIVE LOGITS

of

1.71

who

1.59

沨

1.41

 musées

1.35

摂取

1.34

 intellectuelle

1.31

batt

1.29

 feltro

1.28

 ميد

1.28

 mereka

1.27

Activations Density 0.005%