INDEX

Explanations

legal citations

The neuron activates on numeric tokens (digits or digit sequences), i.e. it detects numbers such as page, volume, and year citations.

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

 these

-1.34

and

-1.32

 this

-1.27

 will

-1.24

GetX

-1.23

-1.18

﹔

-1.16

-1.13

couts

-1.10

将会

-1.09

POSITIVE LOGITS

Adjacent

1.30

ЖА

1.27

だから

1.22

CONCLUSIONS

1.16

 зависит

1.13

 uLocal

1.13

cuz

1.12

 czarna

1.10

çok

1.09

纨

1.09

Activations Density 0.014%