INDEX

Explanations

start of sentence

The neuron is primarily detecting sentence boundaries—especially the punctuation mark ending a sentence and the capitalized token that begins the next one.

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

 then

-1.52

and

-1.48

 will

-1.07

now

-1.05

one

-1.04

in

-0.96

続いて

-0.93

 continue

-0.93

ช่วย

-0.93

Пока

-0.90

POSITIVE LOGITS

 ابتدا

1.60

 první

1.26

 eerst

1.18

 zunächst

1.17

 início

1.16

æde

1.13

Zdrav

1.13

нача

1.12

 estacionamiento

1.12

 شروع

1.10

Activations Density 0.136%