INDEX

Explanations

staff regarding rules or guidelines

The neuron strongly activates on numeric tokens and numeric‐letter combinations used in legal or regulatory citations (e.g. section numbers, rule identifiers, CFR references).

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

ització

-1.24

哪怕

-1.20

 wasn

-1.19

鳎

-1.16

టి

-1.15

⎆

-1.13

 bestemt

-1.13

⟣

-1.13

urture

-1.13

 كما

-1.12

POSITIVE LOGITS

 them

1.48

 available

1.34

 mereka

1.32

 those

1.31

 would

1.30

 они

1.20

 these

1.18

if

1.18

 CONDU

1.17

 they

1.16

Activations Density 0.036%