INDEX

Explanations

fluctuations of the

The neuron selectively activates on numerical tokens (e.g. multi-digit figures, measurements, counts, or statistical values).

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

ఝ

-1.45

ឪ

-1.34

 somente

-1.31

ඍ

-1.30

 breathtaking

-1.30

皷

-1.30

 alcune

-1.30

-1.29

 astounding

-1.28

hicles

-1.26

POSITIVE LOGITS

in

2.00

all

1.83

 each

1.80

on

1.71

for

1.59

 during

1.56

 these

1.53

 with

1.52

 both

1.52

at

1.47

Activations Density 0.104%