INDEX

Explanations

section headers

The neuron detects section-heading labels (e.g. “Objectives,” “Methods,” “Introduction,” “Purpose,” etc.) in academic or scientific documents.

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

ॄ

-1.02

今日も

-1.00

ryk

-0.91

emptory

-0.90

ay

-0.88

heid

-0.88

estría

-0.87

 sowie

-0.87

ßte

-0.86

bym

-0.86

POSITIVE LOGITS

1.37

 several

1.09

1.00

 कई

0.99

 것은

0.95

kela

0.94

 Wartawan

0.93

 many

0.91

duga

0.90

évêque

0.86

Activations Density 0.036%