INDEX

Explanations

figure and section references

The neuron activates on numbered references and labels—e.g. section, proposition, figure, and other numeric identifiers 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

how

-1.64

 obviously

-1.48

any

-1.39

 what

-1.38

 like

-1.35

 much

-1.34

 Schluß

-1.30

 exactly

-1.27

 collezione

-1.26

 more

-1.23

POSITIVE LOGITS

–

1.30

)$,

1.25

笑的

1.23

FunctionFlags

1.23

人了

1.21

"))

1.20

")))

1.19

 <=",

1.17

 brille

1.16

 Кубка

1.16

Activations Density 0.064%