INDEX

Explanations

its and

The neuron fires on discourse markers that signal explanations or overviews (e.g. words like “discuss,” “overview,” “knowledge,” “monitoring,” and “offered”) in instructional or technical text.

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

Ѕ

-1.88

</h1>

-1.76

)));

-1.71

३

-1.71

<b>

-1.69

<h2>

-1.68

-1.65

 These

-1.64

orrhage

-1.60

</b>

-1.58

POSITIVE LOGITS

how

3.25

 what

2.44

why

2.42

its

2.33

 排骨

2.02

 事例

1.83

ᒍ

1.82

 Avez

1.80

 Embaj

1.77

豨

1.74

Activations Density 0.109%