INDEX

Explanations

either ... or choices

This neuron fires on the key technical or formal action verbs (like “requires,” “reported,” “produced,” “opposed,” “use”) that structure academic or scientific statements.

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

܆

-1.76

];

-1.76

 atmosfera

-1.60

na

-1.53

......”

-1.50

urés

-1.48

踽

-1.48

 pengar

-1.47

鰱

-1.46

");

-1.44

POSITIVE LOGITS

or

3.58

you

2.23

timberland

1.88

apnews

1.74

piram

1.73

levis

1.69

nivea

1.67

asics

1.66

tapis

1.65

grats

1.64

Activations Density 0.075%