INDEX

Explanations

to

This neuron turns on for ordinary content words (especially longer or more “academic” terms) in the main expository/argumentative paragraph, as opposed to boilerplate, headlines, or short promotional lines.

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

 rendezvous

0.61

 samostat

0.61

 walkthrough

0.59

 بندر

0.59

 exploratory

0.59

lynk

0.57

 disbursements

0.57

精彩

0.57

 wildcard

0.57

噸

0.57

POSITIVE LOGITS

 patriarchal

1.51

 stereotypes

1.48

 society

1.46

 societal

1.42

传统的

1.40

society

1.40

 stereotypical

1.38

 тради

1.36

传统

1.35

 culturally

1.34

Activations Density 0.826%