INDEX

Explanations

each or every time period

The neuron fires on numeric tokens (digits and numbers), especially counts or measurements 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

楽しめます

-1.46

側は

-1.45

低声

-1.44

 recoge

-1.40

有什么

-1.36

唏

-1.32

 llegará

-1.31

 ayudará

-1.31

зык

-1.30

 obtend

-1.29

POSITIVE LOGITS

for

1.70

ta

1.37

”、“

1.34

 also

1.33

：“

1.31

 focusing

1.30

,”

1.27

遢

1.26

↵

1.24

 نیز

1.24

Activations Density 0.020%