INDEX

Explanations

scientific inquiry nouns followed by prepositions

The neuron fires on high‐level scientific keywords used in paper titles or abstracts to introduce the study focus—words like “effect,” “influence,” “origin,” “model,” “evaluation,” “control,” 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

 أيض

-1.04

 oamenii

-1.02

 vecchia

-1.02

órica

-0.99

があるので

-0.98

 男生

-0.98

èvent

-0.96

 Dica

-0.95

 storica

-0.94

 folosi

-0.94

POSITIVE LOGITS

of

3.42

to

1.24

 about

1.23

"");

1.16

 של

1.08

brand

1.07

 duele

1.06

for

1.03

对此

0.98

 велосипе

0.98

Activations Density 0.235%