INDEX

Explanations

Dutch words and phrases

The neuron selectively fires on tokens that are part of structured metadata or catalog‐style headings—especially Dutch field names, dates, and archival labels.

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

<bos>

-1.48

↵↵

-1.38

 cuando

-1.34

 sommige

-1.32

if

-1.29

̿

-1.25

for

-1.20

 beberapa

-1.19

 которая

-1.19

 если

-1.16

POSITIVE LOGITS

 páss

1.30

cipais

1.28

Źródło

1.28

;";

1.26

 adesso

1.25

頜

1.24

 Eigentü

1.22

 Menjadi

1.22

ContentAsync

1.21

onaise

1.20

Activations Density 0.012%