INDEX

Explanations

multilingual text

The neuron detects major clause boundaries and discourse transitions—especially punctuation like em-dashes, commas, and periods, as well as leading transition words (e.g. “But,” “However,” “While”).

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

Though

-1.43

although

-1.13

Whilst

-1.13

 Though

-1.08

Although

-1.02

 though

-1.00

尽管

-0.95

 Whilst

-0.91

though

-0.90

 jedoch

-0.90

POSITIVE LOGITS

 elevada

1.02

Бележки

1.01

 acolo

1.00

 vielfach

1.00

 Dumnezeu

0.99

Preliminaries

0.98

␻

0.98

マンド

0.98

utente

0.95

 consape

0.95

Activations Density 0.039%