INDEX

Explanations

code structure identifiers

The neuron is detecting numeric literal tokens (whole numbers or decimals) in the text.

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

詬

-1.60

-1.52

ou

-1.42

yn

-1.40

utiny

-1.37

-1.36

 любую

-1.36

},

-1.35

op

-1.33

POSITIVE LOGITS

gruss

1.47

 verhind

1.45

 such

1.43

1.41

 برانيه

1.40

 menyukai

1.38

葑

1.35

 each

1.32

：……

1.32

tapis

1.31

Activations Density 0.177%