INDEX

Explanations

as of a date

The neuron strongly activates on numeric tokens—especially statistics, percentages, and decimal numbers.

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

 during

-1.63

 from

-1.56

 once

-1.55

 when

-1.49

 later

-1.48

当初

-1.45

 today

-1.39

 while

-1.32

 along

-1.31

ğunu

-1.27

POSITIVE LOGITS

حتی

1.29

lain

1.25

 даже

1.23

anything

1.23

她們

1.22

 bahkan

1.20

給我

1.20

所需的

1.16

 حتی

1.16

itself

1.16

Activations Density 0.071%