INDEX

Explanations

Hats, Hatha, Haiti, Red Hat

The neuron flags occurrences of the substring “Hat” (i.e. tokens containing the exact sequence H-a-t).

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

诹

-2.89

羂

-2.83

ௗ

-2.81

美国的

-2.78

 philosophie

-2.69

墬

-2.64



-2.64

 cardiaque

-2.64

齶

-2.64

']

-2.63

POSITIVE LOGITS

</h1>

3.33

</strong>

3.30

The

3.16

 landmark

2.63

2.61

’

2.41

 realizando

2.39

ANY

2.36

氿

2.33

But

2.27

Activations Density 0.006%