INDEX

Explanations

Q: prefix or identifier

The neuron is picking out formatting‐style tokens (especially full-width spaces, bullet dots, and similar punctuation/whitespace markers).

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

necessarily

-1.48

launched

-1.38

ſtre

-1.38

]")]

-1.35

tells

-1.34

conducted

-1.31

decoración

-1.30

')):

-1.29

LECTIONS

-1.29

something

-1.29

POSITIVE LOGITS

Each

1.67

1.65

after

1.45

the

1.41

At

1.41

During

1.39

Before

1.36

otriva

1.36

There

1.32

When

1.32

Activations Density 0.007%