INDEX

Explanations

top ranked lists

The neuron fires strongly on the key words and numbers that appear in list‐style article titles (e.g. “Top 10,” “Best Beaches in America,” “Greatest Players of All Time,” etc.).

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

or

-1.51

 more

-1.32

-1.24

 other

-1.17

 another

-1.12

‐

-1.08

-1.05

\\

-1.04

其他

-1.04

``

-1.02

POSITIVE LOGITS

for

1.38

 которые

1.18

according

1.16

 devriez

1.05

are

1.05

 καλύτε

1.05

 podľa

1.03

 Reihenfolge

1.02

 έχουν

1.02

かを

1.02

Activations Density 0.038%