INDEX

Explanations

reprinted, Mitsuko, Moreover, extend, につき

The neuron consistently activates on the word “reprinted,” i.e. instances noting that material has been or may be reprinted.

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

 ponen

-2.25

selbe

-2.23

者的

-2.20

 Они

-2.08

 participar

-2.06

䊐

-2.03

 zlep

-1.99

 ktore

-1.99

 Шаг

-1.98

幢

-1.98

POSITIVE LOGITS

しても

2.52

鋨

2.16

they

2.02

 olivat

2.00

—

1.94

 There

1.91

 dito

1.91

ヒト

1.88

㐂

1.87

また

1.85

Activations Density 0.000%