INDEX

Explanations

numbers and Roman numerals

The neuron fires on numeric tokens—numbers, Roman numerals, or numbered references (e.g. section/figure/table numbers).

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

â

-1.06

 Usar

-0.98

 kesulitan

-0.94

Organisator

-0.93

avance

-0.93

又能

-0.93

-0.92

 började

-0.91

才可以

-0.90

 aima

-0.90

POSITIVE LOGITS

}%

0.95

 sends

0.84

 espelho

0.83

想起

0.82

'])

0.82

 типов

0.81

ycznie

0.79

]}

0.79

 snippetHide

0.79

 الوزراء

0.78

Activations Density 0.054%