INDEX

Explanations

words or labels

The neuron selectively flags spans of formatted label or sign–style text, especially words or short phrases rendered in all caps (often quoted) like “DORK MAGNET,” RAILROAD CROSSING, DONATING, FRANKLIN D. ROOSEVELT, 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

 levure

-1.21

rags

-1.16

 tasche

-1.16

 SANTOS

-1.14

 chirurgie

-1.13

 folgenden

-1.13

になるので

-1.13

magine

-1.13

FONTE

-1.12

 saisir

-1.11

POSITIVE LOGITS

if

1.43

你不要

1.29

for

1.27

你们的

1.09

1.08

1.07

 Mitarbeitern

1.07

1.06

and

1.05

if

1.05

Activations Density 0.171%