INDEX

Explanations

conjunctions and transitions

The neuron is essentially firing on semicolons (the “;” token) as clause‐ or list‐separators in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 Και

0.87

은

0.83

ค

0.82

ค

0.82

는

0.79

 รอ

0.79

 ทั้ง

0.77

 ไป

0.76

?)

0.75

Τ

0.75

POSITIVE LOGITS

 inoltre

1.60

此外

1.35

 moreover

1.34

 furthermore

1.34

 zudem

1.33

 additionally

1.29

 همچنین

1.27

 এছাড়াও

1.24

 außerdem

1.23

然而

1.21

Activations Density 0.301%