INDEX

Explanations

contrasting conjunctions

The neuron strongly activates on stative or linking verbs (e.g. “is,” “was,” “retains,” “appears,” “lived,” “provides”) that express states or properties.

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

或者是

0.80

but

0.71

したがって

0.68

 therefore

0.67

 поэтому

0.65

 więc

0.64

 dunque

0.63

name

0.62

 donc

0.62

Therefore

0.62

POSITIVE LOGITS

 nonetheless

1.59

 dennoch

1.45

 nevertheless

1.43

 Nonetheless

1.37

 Nevertheless

1.30

 trotzdem

1.27

Nonetheless

1.21

Nevertheless

1.16

 vẫn

1.13

 toch

1.13

Activations Density 1.299%