INDEX

Explanations

question or list item starters

The neuron fires primarily on numerals (e.g. question numbers, line numbers, and other standalone numeric tokens).

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

 diamants

-1.54

――――

-1.48

 aspek

-1.34

>?

-1.34

 radikal

-1.30

suchen

-1.26

 klinik

-1.26

 superbes

-1.22

 konflik

-1.22

 kehadiran

-1.21

POSITIVE LOGITS

how

2.88

why

2.44

 what

2.05

 describe

1.91

How

1.76

 What

1.51

 explain

1.51

how

1.50

who

1.48

can

1.45

Activations Density 0.156%