INDEX

Explanations

specifies time or key

The neuron fires on numeric literals (tokens containing digits) in code or configuration text.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 駅前

-1.53

かを

-1.41

んだん

-1.34

 två

-1.33

pullover

-1.30

 بدأ

-1.27

 estran

-1.25

 hintergrund

-1.25

sehr

-1.21

bảng

-1.21

POSITIVE LOGITS

 your

1.84

or

1.81

 there

1.74

if

1.73

you

1.66

杈

1.63

tized

1.53

ֶּ

1.51

͝

1.48

alties

1.47

Activations Density 0.040%