INDEX

Explanations

list item identifiers

This neuron activates on numeric tokens and list‐item labels (e.g. numbers, “a”, “b”, “i”, “ii”) in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 agres

-1.16

 behar

-1.16

 Terima

-1.16

así

-1.13

転車

-1.13

eload

-1.13

 Meksi

-1.11

Interes

-1.11

}?>

-1.09

 proiect

-1.09

POSITIVE LOGITS

you

1.41

If

1.32

use

1.27

psies

1.20

 asked

1.16

let

1.15

 National

1.15

 make

1.15

You

1.14

 using

1.14

Activations Density 0.008%