INDEX

Explanations

contradictions

The neuron activates on high-impact marketing or promotional claim phrases—i.e. strong sales-pitch statements.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

another

0.59

近年

0.57

recent

0.52

近年来

0.51

pubmed

0.50

imuth

0.50

 récent

0.49

 epigenetic

0.49

 linkages

0.49

 récente

0.49

POSITIVE LOGITS

 That

1.14

 This

1.11

 After

1.11

 Afterward

1.11

 When

1.05

 Then

1.04

 Although

1.03

 Because

1.01

And

1.00

 Consequently

0.99

Activations Density 0.399%